diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--84xaxhg.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--84xaxhg.json
new file mode 100644
index 0000000000000000000000000000000000000000..ab06b210a588bfc79f10ab02cd193b1d46471b9f
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--84xaxhg.json
@@ -0,0 +1,222 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-privileged--84xaxhg",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 990,
+   "sim_steps": 990,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 464,
+   "agent_resets": 26453,
+   "agent_sim_steps": 10412,
+   "agent_records": 0,
+   "agent_replays": 10,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 18.162,
+   "peak_processes": 32,
+   "peak_disk_gb": 0.987,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-privileged--84xaxhg/replay.mp4",
+  "duration_s": 39.64,
+  "fps": 25.0,
+  "frames": 991,
+  "sha256": "dc36673849e3ca9e38f112b0bb59f7ce43da1f0b639348b1b8f1170344509d3a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 19,
+      "text_zh": "观察：log step 19 所读模型将 ball1 绑定黄色材质、ball2 绑定粉色材质；本次视频 0.0 s 右侧近景可见黄色球在掌内靠拇指侧，粉色球靠其外侧，以下按此固定身份追踪，不按变色验收。"
+     },
+     {
+      "t": 4.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0–9.5 s 每隔约0.5 s 的近景显示，MyoHand起初轻微屈指、两球沉降，随后黄色球基本停在拇指与掌面之间；小目标标记继续绕行，视频 4.0–6.5 s 可见标记移到粉色球一侧而黄色球未随行。"
+     },
+     {
+      "t": 39.6,
+      "log_step": null,
+      "text_zh": "推断：视频 10.0–39.0 s 逐秒近景及视频 38.6–39.6 s 末段显示黄色球始终留在同一掌内区域，无持续绕行，因此不存在连续200个控制步的完整目标跟踪窗口，终态也未满足；结论依据球与移动标记的相对运动，而非仅由总评分为零推得。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频 0.0–2.0 s MyoHand屈指、掌面和拇指侧承托黄色球1；视频 2.0–39.6 s 球仍由同一手的指掌承托，未见持续滚动换位。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.0,
+      "log_step": 19,
+      "text_zh": "观察：log step 19 的模型材质定义将粉色球标为 ball2；视频 0.0–9.5 s 的密集近景中，粉色球起初随手部沉降，随后停在指根附近。视频 4.0–7.5 s 的小目标标记从球旁经过、跨过其投影，粉色球未跟着标记在掌内绕转。"
+     },
+     {
+      "t": 39.6,
+      "log_step": null,
+      "text_zh": "推断：视频 10.0–39.0 s 的逐秒画面与视频 38.6–39.6 s 的密集末段中，粉色球一直位于黄色球的同一侧，缺少连续一整圈所需的换位，故球2持续跟踪节点从未达到且终态为假；球或标记的颜色变化不足以替代三维距离判定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频 0.0–2.0 s 粉色球2随MyoHand指掌姿态变化落稳在指根附近，视频 2.0–39.6 s 持续由同一手的指掌承托，未见将其绕掌输送的手内调整。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 2.0,
+   "log_step": 189,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 2.0 s MyoHand已将两球托稳，随后视频 2.0–9.5 s 目标标记绕行而球1停在掌内；这一偏离持续到视频 39.6 s，球2也并行未能持续跟踪。log step 189 将最终交付改为恒定肌肉命令，log step 202 对该990行文件的两次回放均为 success=false、instance_ok=true、deterministic=true；断点以并行失败的球1跟踪步骤为代表，未把此前调试方案的掉球算入本次轨迹。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 4.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 2.0–9.5 s，MyoHand保持近似固定托球姿势，两球中心基本不换位，小目标标记却持续绕行；视频 4.0–6.5 s 标记经过粉色球附近而黄色球仍停在拇指侧，之后视频 10.0–39.6 s 重复这一停球、走标记的状态。机器人没有跟上移动目标。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 189 的最终生成脚本默认用 np.tile(u0, (T, 1)) 重复同一肌肉命令，替换了未能稳定运行的脉冲控制；log step 202 重新生成并两次回放验证的是这一恒定命令文件，均未成功。视频 2.0–39.6 s 对应表现为托稳两球但不使其持续换位，说明交付的肌肉控制只能维持托持，未实现任务所需的掌内绕转；现有证据不足以细分为接触几何、逆解或反馈参数问题。"
+   }
+  ],
+  "gap_zh": "观察：本目录有完整视频与运行日志，但没有逐控制步状态日志；run.json 的 reward 未报告 target1_dist、target2_dist 或 solved_run，故三项原生指标均填 null。目标标记很小且部分被指掌和球遮挡，无法恢复各球精确三维误差及瞬间合格窗口；两球缺少整圈换位仍足以判定持续跟踪未完成。log step 203 所称约19步合格是agent自述，未作为节点到达时间；其“不可达”判断也未作为任务或环境根因。",
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  }
+ },
+ "summary_zh": "观察：本次视频 0.0–39.6 s 中MyoHand将双球托在掌内，起初轻微沉降后基本停位，两项持续跟踪节点均未完成；最终文件在 log step 202 两次一致回放中均失败。推断：断点取并行失败的球1持续跟踪（操作中），表现为双球未跟上移动目标，交付的恒定肌肉命令未形成持续掌内绕转；缺少逐步三维误差，不能细分控制根因或精确计量短暂合格窗口。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "两节点均未到达、final=false 与视频一致（球全程停在掌内，无绕转）；断点、现象、根因有据，未见需改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:02:32Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--84xaxhg/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--84xaxhg/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c536c79b1d916efc3e94db1e799b1464a66ef7959b9cdc242c1765884f4fb829",
+   "video_sha256": "dc36673849e3ca9e38f112b0bb59f7ce43da1f0b639348b1b8f1170344509d3a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--aab3yvm.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--aab3yvm.json
new file mode 100644
index 0000000000000000000000000000000000000000..0ae038c704c07158f8de8d838a1faf1ccec50d59
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--aab3yvm.json
@@ -0,0 +1,225 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-privileged--aab3yvm",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 460,
+   "sim_steps": 460,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 194,
+   "agent_resets": 13859,
+   "agent_sim_steps": 12583,
+   "agent_records": 44,
+   "agent_replays": 47,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 4.731,
+   "peak_processes": 14,
+   "peak_disk_gb": 0.266,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-privileged--aab3yvm/replay.mp4",
+  "duration_s": 18.44,
+  "fps": 25.0,
+  "frames": 461,
+  "sha256": "0076cc5c7027a247666864a1df832a4ca375a7ede879f8da29f7b5e94a0429a6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–7.75 s 的右侧近景中，MyoHand调整腕部及手指后，两球仍保持一球在画面左上、另一球在右下的相对布局，仅有小幅位移；小目标标记继续绕行，两球没有随之绕转换位。对object1的持续跟踪不能由短暂靠近或球体变色认定。"
+     },
+     {
+      "t": 18.4,
+      "log_step": null,
+      "text_zh": "推断：视频 8.00–18.40 s（含视频 17.40、17.60、17.80、18.00、18.20、18.40 s 末段近景）仍显示相同的双球布局，目标标记周期性移到球外侧；相比成功参考视频中双球随目标交换位置，本次object1没有覆盖完整目标绕行所需的连续200步跟踪窗口，因此本节点从未成立且终值为false。此判断基于运动不匹配，不是仅由总评分为0反推。"
+     },
+     {
+      "t": 18.4,
+      "log_step": 376,
+      "text_zh": "观察：log step 374恢复最终460步交付文件，log step 376对该文件的Session.replay返回steps=460、success=false、done=false、solved_run=0、hold_steps=200；log step 377官方评分器两次回放均success=false且终态哈希相同。共同计数只支持整项未完成，不单独替代逐球视频判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.00–18.40 s，MyoHand同一只手的手指及掌部持续接触并留住第1颗球，伴随腕指小幅运动在手内调整球的位置；未见计划松手，实际未形成随目标绕转换位。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–7.75 s 的右侧近景中，MyoHand调整腕部及手指后，两球仍保持一球在画面左上、另一球在右下的相对布局，仅有小幅位移；小目标标记继续绕行，两球没有随之绕转换位。对object2的持续跟踪不能由短暂靠近或球体变色认定。"
+     },
+     {
+      "t": 18.4,
+      "log_step": null,
+      "text_zh": "推断：视频 8.00–18.40 s（含视频 17.40、17.60、17.80、18.00、18.20、18.40 s 末段近景）仍显示相同的双球布局，目标标记周期性移到球外侧；相比成功参考视频中双球随目标交换位置，本次object2没有覆盖完整目标绕行所需的连续200步跟踪窗口，因此本节点从未成立且终值为false。此判断基于运动不匹配，不是仅由总评分为0反推。"
+     },
+     {
+      "t": 18.4,
+      "log_step": 376,
+      "text_zh": "观察：log step 374恢复最终460步交付文件，log step 376对该文件的Session.replay返回steps=460、success=false、done=false、solved_run=0、hold_steps=200；log step 377官方评分器两次回放均success=false且终态哈希相同。共同计数只支持整项未完成，不单独替代逐球视频判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.00–18.40 s，MyoHand同一只手的手指及掌部持续接触并留住第2颗球，伴随腕指小幅运动在手内调整球的位置；未见计划松手，实际未形成随目标绕转换位。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": 0
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 1.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.75–2.00 s，机器人已开始腕指操作并留住双球，但球主要停在手中原来的相对位置，未随继续绕行的目标换位；视频 2.00–18.40 s 未修复这种运动不匹配。两节点并行操作且均未完成，此处以球1持续跟踪作为共同跟踪失败的断点锚点，不表示球2先完成。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 1.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 1.00–7.75 s及视频 8.00–18.40 s，MyoHand上的双球基本维持并排布局，而小目标标记持续绕行并反复离开球心所在位置；机器人没有让双球跟上移动目标。视频末段双球仍在手上，没有可见掉落，失败表现为未跟上目标运动。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 321的生成脚本用周期函数设定腕部、拇指及其余手指关节目标，再以PD力矩和肌肉力矩映射求肌肉命令；log step 374、376–377确认最终选取并重复回放的是460步肌肉命令文件。结合视频 1.00–18.40 s中双球被留在手上却未随目标绕行，说明此次腕指协调控制未产生所需的球体跟踪运动；现有记录不足以定位到某个控制参数、接触几何或逆解子问题。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志；最终交付文件的回放记录（log step 376–377）未报告两颗球各自的距离历史，视频透视及局部遮挡也不能精确量出三维0.015米误差或阈值越界时刻。可由全段缺少绕转换位判定两节点均未形成连续200步窗口，但不能细分控制根因；native_metrics的两个距离值因此为null，solved_run=0取自log step 376的最终文件回放结果。"
+ },
+ "summary_zh": "本次460步评分轨迹中，MyoHand持续留住双球并进行小幅腕指调整，但双球基本维持并排布局，没有像成功参考视频那样随固定配对目标绕转换位，两个持续跟踪节点均未完成。断点记为球1持续跟踪的操作中，现象为未跟上移动目标，根因归入控制；log step 376–377确认最终交付回放失败且可确定复现。缺少逐球距离和保持历史，无法精确定位阈值越界时刻或进一步细分控制原因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "两节点未成立、终值false与视频一致（双球全程停在手上、布局基本不变，无掉落），断点、动态时机现象和控制根因（停在父类）有证据支持，未发现需要改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:56:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--aab3yvm/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--aab3yvm/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "863deb7a904d1b7acab8896b634ff18067a6ebb6c40121c8836b631e73010c29",
+   "video_sha256": "0076cc5c7027a247666864a1df832a4ca375a7ede879f8da29f7b5e94a0429a6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--pqfxjpu.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--pqfxjpu.json
new file mode 100644
index 0000000000000000000000000000000000000000..bf5a272f83e0a1dc01ac40111a2a7d14924067e6
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--pqfxjpu.json
@@ -0,0 +1,237 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-privileged--pqfxjpu",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 155,
+   "sim_steps": 155,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 54,
+   "agent_resets": 276,
+   "agent_sim_steps": 4921,
+   "agent_records": 0,
+   "agent_replays": 3,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-privileged--pqfxjpu/replay.mp4",
+  "duration_s": 6.24,
+  "fps": 25.0,
+  "frames": 156,
+  "sha256": "2172bd028f4f50589dbc4ca1cb14875291446f9035171b8a2b3483588cc30862"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.64,
+      "log_step": 18,
+      "text_zh": "观察：视频0.00–1.64 s，MyoHand屈伸手指、转动手腕，双球留在指掌接触区。log step 18的所交付轨迹生成记录在循环索引40报告球1误差0.0077 m、球2误差0.0111 m，共同连续合格计数36；这只是短时合格，尚未达到200步。"
+     },
+     {
+      "t": 2.04,
+      "log_step": 18,
+      "text_zh": "推断：log step 18的循环索引50是第51次控制，映射视频51/25=2.04 s；该处报告球1误差0.0157 m，之后抽样值持续超阈，至循环索引140为0.0768 m。视频2.04、2.44、3.24、4.04 s可见手指继续调整球；这些动作未使球1持续跟上移动目标。上述映射已用第155次控制掉球对应视频6.20 s末帧核对，不使用agent工作时间。"
+     },
+     {
+      "t": 6.2,
+      "log_step": 30,
+      "text_zh": "推断：评分器字段n_actions=sim_steps=155，log step 23保存的155行轨迹在log step 30两次回放均于第155个控制步失败、终态哈希一致；录像共156帧（含初态），末帧视频6.20 s可见掉球，与此吻合。每控制步对应一帧，155步不足节点要求的连续200步，因此本回合该节点从未成立，不能把短暂贴近目标记为完成。"
+     },
+     {
+      "t": 6.2,
+      "log_step": 18,
+      "text_zh": "观察：视频5.92–6.20 s，双球向指掌支承区外滑离，左侧视角可见它们与手明显分开。log step 18的末次循环索引154报告球1高度1.224 m、误差0.2609 m，低于掉球高度1.25 m；其末态哈希与log step 23、30的交付回放相同。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "视频0.00–5.92 s可见单只MyoHand的手指屈伸、配合掌面接触并调整掌中球1的位置，末段球滑离支承区。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.64,
+      "log_step": 18,
+      "text_zh": "观察：视频0.00–1.64 s，MyoHand在同一只手的指掌间同时调整两球。log step 18对应交付轨迹的循环索引40报告球2误差0.0111 m、共同连续合格计数36，尚未达到200步。"
+     },
+     {
+      "t": 2.04,
+      "log_step": 18,
+      "text_zh": "观察：log step 18在循环索引50报告球2误差0.0185 m；循环索引70、80又分别为0.0132、0.0128 m，但末次循环索引154为0.1226 m、高度1.343 m。视频2.04–5.92 s手指继续在掌中调整双球，视频6.00–6.20 s可见双球滑出指掌支承区；未将球的颜色作为球2自身距离达标的证据。"
+     },
+     {
+      "t": 6.2,
+      "log_step": 30,
+      "text_zh": "推断：评分器字段n_actions=sim_steps=155，log step 23保存的155行轨迹在log step 30两次回放均于第155个控制步失败、终态哈希一致；录像共156帧（含初态），末帧视频6.20 s可见掉球，与此吻合。每控制步对应一帧，155步不足节点要求的连续200步，因此本回合该节点从未成立，不能把短暂贴近目标记为完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "视频0.00–5.92 s可见同一只MyoHand以屈伸的手指和掌面同时接触、滚动调整球2，末段未能留住球。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 2.04,
+   "log_step": 18,
+   "where": "操作中",
+   "evidence_zh": "推断：MyoHand正在并行调整双球以跟踪移动目标。log step 18对应交付轨迹在循环索引40尚报告两球误差合格，循环索引50球1误差已为0.0157 m，此后抽样误差持续增大，末次球1高度降至1.224 m；log step 23保存该轨迹，log step 30两次回放均在155步失败且终态哈希一致。视频1.60–2.40 s可见手指持续调整但没有完成绕转，视频5.92–6.20 s可见掉球，故将球1持续跟踪操作中的未修复失跟定位为断点，视频2.04 s为粗略导航锚点。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 2.04,
+    "log_step": 18,
+    "evidence_zh": "推断：移动目标继续绕行时，MyoHand未能让球1同步跟上。log step 18记录循环索引40球1误差0.0077 m，循环索引50为0.0157 m，循环索引80为0.0306 m、100为0.0409 m；对应视频1.64、2.04、3.24、4.04 s，手指仍在调整球，球集中于指掌一侧，没有随目标完成持续绕转。此处是运动目标的持续失跟。"
+   },
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 6.0,
+    "log_step": 18,
+    "evidence_zh": "观察：视频5.20–5.92 s双球仍处指掌附近，视频6.00–6.20 s它们滑离原支承区并与手分开，MyoHand没有重新留住球。log step 18末次记录球1高度1.224 m低于1.25 m，log step 23、30对所交付文件的回放均报告掉球失败；这发生在继续掌内操作时。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 16–18的脚本以两球目标误差为代价搜索手部关节设定，再用肌肉控制层执行；log step 18显示交付轨迹在短时合格后误差持续扩大，log step 23、30确认原执行与交付回放终态一致。视频2.04–6.20 s中手部持续动作却未维持双球绕行，最终失去支承，支持肌肉与手指协调控制不足；现有证据不能把它进一步归为特定接触布局、反馈律或参数错误。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志，log step 18仅每隔若干控制步输出误差，无法确定两颗球首次越阈的精确先后或重建所有瞬时接触；断点以球1持续失跟并最终掉落为依据。小目标标记受手指及球遮挡，二维画面不能独立测定0.015 m三维阈值，球2变色也不能证明自身距离合格。评分器reward未单列target1_dist、target2_dist、solved_run，原生指标按要求保留null，交付执行日志的数值仅写入证据；根因停留在控制大类。"
+ },
+ "summary_zh": "MyoHand确实通过指掌接触同时调整双球，但交付只有155个控制步，两个要求连续200步的跟踪节点均未达成，终态为0/2。球1在视频约2.04 s已失跟且未完成修复，视频6.00–6.20 s双球滑离支承区，球1掉落触发失败；交付回放一致，证据支持控制执行不足。缺少逐控制步状态及评分器原生距离字段，无法精确区分两球首次越阈的先后与更细控制根因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "两节点均未达成（0/2）、时钟映射（控制步/25）与末帧掉球视频6.20 s吻合，断点在球1失跟、现象与根因均有log step 18/23/30及视频支持，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:00:44Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--pqfxjpu/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--pqfxjpu/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "55fb6b1c6c05182df464977a31cb7d10d9d2da4942fe4a31693bebe57bbae8eb",
+   "video_sha256": "2172bd028f4f50589dbc4ca1cb14875291446f9035171b8a2b3483588cc30862",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--rrazfuz.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--rrazfuz.json
new file mode 100644
index 0000000000000000000000000000000000000000..64dc81e9f86c68be73eaf94b4aaf3b7b175cd244
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--rrazfuz.json
@@ -0,0 +1,218 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-privileged--rrazfuz",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 400,
+   "sim_steps": 400,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 47,
+   "agent_resets": 9576,
+   "agent_sim_steps": 7074,
+   "agent_records": 9280,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.646,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.837,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 1,
+   "session_log_from_monitor": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-privileged--rrazfuz/replay.mp4",
+  "duration_s": 16.04,
+  "fps": 25.0,
+  "frames": 401,
+  "sha256": "77ff87080f9802e303bfe6c99c7cef4d8781812b7e0b79dc658f8f930483d197"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.0,
+      "log_step": 206,
+      "text_zh": "观察：视频 0.0–16.0 s 的两视角及右侧近景加密帧显示，MyoHand屈指承托两球并摆动腕部，但两球在掌内大体保持原来的相对位置；视频 2.0–4.0 s、10.0–12.0 s 可见小目标标记绕至球侧而球未随之绕转换位。log step 205 保存ES9检查点生成的400步动作；log step 206 对该次生成记录的球1目标距离，在控制步100、150、200、250、300、350、400分别为0.0454、0.0442、0.0181、0.0200、0.0457、0.0446、0.0179米，全部高于0.015米。"
+     },
+     {
+      "t": 16.0,
+      "log_step": 206,
+      "text_zh": "推断：交付记录（log step 205–206）与评分器n_actions=sim_steps=400、deterministic=1一致；视频含初始帧及400个动作帧，25 fps，故控制步100/25对应视频 4.0 s、控制步400/25对应视频 16.0 s，两处均已核对球仍在掌中且与小目标标记相离。上述每隔50控制步的越界采样排除了全程任何连续200步合格窗口，因此球1持续跟踪从未到达、终态为false；不由总失败分数或球体变色单独反推。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.0–0.6 s、4.8–6.4 s、12.8–16.0 s，单只MyoHand的屈曲手指配合掌面接触并约束掌中object1，随腕部摆动做小幅滚动、换位；没有把球松开后重新抓取，实际调整未形成所需持续绕转。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.0,
+      "log_step": 206,
+      "text_zh": "观察：视频 0.0–16.0 s 的两视角及右侧近景加密帧显示，MyoHand屈指承托两球并摆动腕部，但两球在掌内大体保持原来的相对位置；视频 2.0–4.0 s、10.0–12.0 s 可见小目标标记绕至球侧而球未随之绕转换位。log step 205 保存ES9检查点生成的400步动作；log step 206 对该次生成记录的球2目标距离，在控制步100、150、200、250、300、350、400分别为0.0470、0.0468、0.0180、0.0207、0.0468、0.0455、0.0166米，全部高于0.015米。"
+     },
+     {
+      "t": 16.0,
+      "log_step": 206,
+      "text_zh": "推断：交付记录（log step 205–206）与评分器n_actions=sim_steps=400、deterministic=1一致；视频含初始帧及400个动作帧，25 fps，故控制步100/25对应视频 4.0 s、控制步400/25对应视频 16.0 s，两处均已核对球仍在掌中且与小目标标记相离。上述每隔50控制步的越界采样排除了全程任何连续200步合格窗口，因此球2持续跟踪从未到达、终态为false；不由总失败分数或球体变色单独反推。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.0–0.6 s、4.8–6.4 s、12.8–16.0 s，单只MyoHand的屈曲手指配合掌面接触并约束掌中object2，随腕部摆动做小幅滚动、换位；没有把球松开后重新抓取，实际调整未形成所需持续绕转。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 2.0,
+   "log_step": 206,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 1.4–4.0 s，MyoHand继续接触承托两球并摆动手腕，小目标标记继续绕行，两球却仍并排停留在原来的指掌区域；视频 10.0–12.0 s 再次出现同样偏离。log step 206 记录最终交付轨迹的控制步100、150两球误差均超过0.04米，最长共同合格仅33控制步。推断：两项并行跟踪均在操作中失败且之后未完成200步窗口，以球1节点作断点导航，球2同时未达，不表示两球有先后顺序。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 2.0,
+    "log_step": 206,
+    "evidence_zh": "观察：视频 1.4–6.4 s 及9.6–14.4 s，MyoHand保持两球在指掌间，两球主要随腕部整体移动，小目标标记则绕到球的侧面、下方再绕回，球没有持续跟上移动标记；视频 15.0–16.0 s 仍是相同掌内布局。log step 206 对最终保存动作记录最长共同合格33控制步、最终solved_run=0。推断：可见缺陷是对周期移动目标的跟随不足，属动态时机。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 205,
+    "evidence_zh": "观察：log step 158 的评估器以两球目标误差、速度、手关节、肌肉激活和目标相位计算肌肉动作；log step 165、173、185、205沿用该评估器并最终保存ES9生成的400步动作。log step 206 在这次生成中记录两球误差周期性增至约0.045米，最长共同合格33控制步；视频 2.0–6.4 s、10.0–14.4 s 对应地显示手腕在动而两球未完成掌内绕转。推断：该肌肉控制策略能保住球，却未产生追随目标所需的连续掌内运动，归为控制；现有证据不能进一步区分接触设计、轨迹与反馈环节。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志或接触力、肌肉响应记录；目标标记较小且部分被球与手指遮挡，不能从二维画面独立复核每步三维距离或精确定位首次越界。节点未达由视频的缺失绕转与最终交付日志的密集越界采样共同确定，精确失败时刻仅用约视频 2.0 s 导航；根因仅判到控制大类。评分器未报告DAG三个原生指标的数值，native_metrics的value均为null。"
+ },
+ "summary_zh": "交付的400控制步轨迹中，MyoHand持续接触承托并小幅调整两球，但两球主要随腕部整体摆动，未随各自移动目标持续绕转，两个跟踪节点均未达到连续200步要求，终态0/2。断点记球1持续跟踪（两节点并行失败），现象为动态时机，根因为控制策略未产生足够掌内运动；log step 206记录最长共同合格仅33控制步，评分器回放失败且未触发掉球终止。缺少逐步状态与接触力记录，不能细分控制根因；成功参考视频中明显的掌内绕转换位仅用于外观比较。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "两个跟踪节点未达、终态0/2与评分器success=0、log step 206的7个采样点（全部>0.015 m，最长共同合格33步）和视频帧一致；断点、现象、根因有证据支持，未发现需要修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:58:26Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--rrazfuz/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--rrazfuz/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "5e1ecc58b95953f1241508ab832ff801b92889448b13a1b144c0b50d16bb82bb",
+   "video_sha256": "77ff87080f9802e303bfe6c99c7cef4d8781812b7e0b79dc658f8f930483d197",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--x2hpb7x.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--x2hpb7x.json
new file mode 100644
index 0000000000000000000000000000000000000000..fe50acd47a9ff412faab9ea874268dcbf1c49be6
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--x2hpb7x.json
@@ -0,0 +1,228 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-privileged--x2hpb7x",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 1000,
+   "sim_steps": 1000,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 62,
+   "agent_resets": 65,
+   "agent_sim_steps": 7166,
+   "agent_records": 0,
+   "agent_replays": 7,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.074,
+   "peak_processes": 15,
+   "peak_disk_gb": 0.291,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-privileged--x2hpb7x/replay.mp4",
+  "duration_s": 40.04,
+  "fps": 25.0,
+  "frames": 1001,
+  "sha256": "c022a85bc94066a0767cad23943c94c17cab6b53fbc744b2758d3a1797f1714a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0–3.0 s，MyoHand收拢手指，两球在指掌间移动；球1（黄色球）随后停留在掌中一侧。起始位置或短暂靠近目标不构成连续200步跟踪。"
+     },
+     {
+      "t": 8.0,
+      "log_step": null,
+      "text_zh": "观察：视频 3.0–8.0 s 的近景及视频 9.0–40.0 s 每秒帧显示，球1仅小幅移动，未随移动标记绕掌持续换位；末秒视频 39.04–40.0 s 仍如此。成功参照视频 0.0–8.0 s 则能看见双球沿指掌持续绕转，该运动线索在本次视频中缺失。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 88,
+      "text_zh": "推断：本次视频可见球1长期近乎停在掌中，结合任务要求跟随绕行目标连续200步，判定该节点未完成。log step 87 保存最终轨迹，log step 88 对该文件两次回放均返回 success=false、steps=1000，原因均为目标轨迹耗尽，与视频一致；未把会话其他候选的数值当作该球的达成记录。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.2–3.0 s，MyoHand单手屈伸手指，配合掌面使球1（黄色球）在手内滚动换位；视频 4.0–40.0 s 主要围托球并作小幅调整，未形成持续绕转。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0–3.0 s，MyoHand收拢手指，两球在指掌间移动；球2（橙色球）随后停留在掌中一侧。起始位置或短暂靠近目标不构成连续200步跟踪。"
+     },
+     {
+      "t": 8.0,
+      "log_step": null,
+      "text_zh": "观察：视频 3.0–8.0 s 的近景及视频 9.0–40.0 s 每秒帧显示，球2仅小幅移动，未随移动标记绕掌持续换位；末秒视频 39.04–40.0 s 仍如此。成功参照视频 0.0–8.0 s 则能看见双球沿指掌持续绕转，该运动线索在本次视频中缺失。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 88,
+      "text_zh": "推断：本次视频可见球2长期近乎停在掌中，结合任务要求跟随绕行目标连续200步，判定该节点未完成。log step 87 保存最终轨迹，log step 88 对该文件两次回放均返回 success=false、steps=1000，原因均为目标轨迹耗尽，与视频一致；未把会话其他候选的数值当作该球的达成记录。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.2–3.0 s，MyoHand单手屈伸手指，配合掌面使球2（橙色球）在手内滚动换位；视频 4.0–40.0 s 主要围托球并作小幅调整，未形成持续绕转。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 3.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 2.0–4.0 s，MyoHand已收拢并接触双球，但球的位置逐渐停在掌中两侧；之后直到视频 40.0 s 都未出现持续绕行。推断：双球两个并行跟踪步骤共同受阻，以球1持续跟踪作为断点定位，球2也未完成；视频 3.0 s 为停滞的粗略导航锚点，不是精确距离越界时刻。log step 88 对最终文件的两次回放均在1000步目标轨迹耗尽时判失败。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 3.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.0–3.0 s 双球随手指动作短暂移位；视频 3.0–8.0 s 及视频 9.0–40.0 s 双球基本保持各自的掌中位置，而小目标标记继续变化位置，球未跟随目标持续绕转。推断：这是对移动目标没有跟上的表现；末秒视频 39.04–40.0 s 仍有手掌承托双球，失败不表现为掉球。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 87,
+    "source_zh": "日志",
+    "evidence_zh": "观察：log step 78 的搜索脚本使用固定姿态、衰减瞬态和周期肌肉命令生成候选动作；log step 83 输出 best5 候选，log step 84–87 将该参数写入生成脚本，修复语法后成功生成并保存最终轨迹。log step 88 对最终文件两次确定性回放均失败；视频 3.0–40.0 s 显示其肌肉命令主要形成静态围托，未产生双球所需的连续手内绕行动作。推断：交付的肌肉控制序列能够留住球，却未实现目标轨迹跟踪；证据不足以进一步区分接触设计、轨迹求解或具体参数问题，停在控制层。"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志，评分器reward也未提供target1_dist、target2_dist和solved_run数值，三个原生指标均填null；不能从二维投影精确恢复三维0.015米误差或某次瞬间越界时刻，也不以球体变色代替验收。两个节点的false依据是全程可见的持续绕行缺失，配合最终文件回放失败，而非仅依据零分；断点两个并行步骤同步受阻，选球1定位不代表已证实它先于球2失败。"
+ },
+ "summary_zh": "MyoHand在评分视频中收拢手指并留住双球，但两球随后基本停在掌中，两个持续跟踪节点均未完成，终态为0/2。断点定位球1持续跟踪的操作中，球2同步未完成；现象是未跟上移动目标，交付的周期肌肉控制序列未能产生持续手内绕转，最终文件两次回放均因1000步目标轨迹耗尽失败（log step 88）。没有逐步三维距离与连续计数记录，不能精确定位阈值越界或进一步细分控制根因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "两节点未完成、断点、现象（动态时机，没跟上）和控制根因与视频及log step 88的两次回放失败一致；未发现需要修改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:56:40Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--x2hpb7x/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--x2hpb7x/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "7a057f0a16ba487103ef01f42e33d30de9d4828055b92b46bbdf9c06645cbcc9",
+   "video_sha256": "c022a85bc94066a0767cad23943c94c17cab6b53fbc744b2758d3a1797f1714a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--y8ssqwd.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--y8ssqwd.json
new file mode 100644
index 0000000000000000000000000000000000000000..374b0eacc3a40956596e7311b90d7e8c817f1067
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-privileged--y8ssqwd.json
@@ -0,0 +1,219 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-privileged--y8ssqwd",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 82,
+   "sim_steps": 82,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 32,
+   "agent_resets": 25738,
+   "agent_sim_steps": 1375,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.833,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.244,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-privileged--y8ssqwd/replay.mp4",
+  "duration_s": 3.32,
+  "fps": 25.0,
+  "frames": 83,
+  "sha256": "b5bcad7b5c6aa879cb36d9af3806a2593b223137da7b5b66994ff5a27eade25a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：视频0.00–3.04 s，MyoHand的手指弯曲，黄色球1与另一球在指掌内移动；视频3.12–3.28 s，两球迅速分离并离开手指围成的支承区域。已检查本次全部83帧，包含最后一秒的逐帧画面。"
+     },
+     {
+      "t": 3.28,
+      "log_step": 63,
+      "text_zh": "推断：评分器reward.n_actions与sim_steps均为82；log step 63中agent脚本明确生成并保存形状为(82,39)的失败轨迹，与本视频的短回合及末段脱手一致。球1所需的连续200控制步窗口不可能在仅82控制步的交付回合内形成，故全程无达标区间、终态为false；此判断不以总分0或球体变色代替判据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.40–3.04 s，MyoHand单手的弯曲手指围住掌内的黄色球1，随手指屈曲使其在指掌间滚动、换位；视频3.12–3.28 s最终脱手。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：视频0.00–3.04 s，MyoHand的手指弯曲，粉褐色球2与另一球在指掌内移动；视频3.12–3.28 s，两球迅速分离并离开手指围成的支承区域。已检查本次全部83帧，包含最后一秒的逐帧画面。"
+     },
+     {
+      "t": 3.28,
+      "log_step": 63,
+      "text_zh": "推断：评分器reward.n_actions与sim_steps均为82；log step 63中agent脚本明确生成并保存形状为(82,39)的失败轨迹，与本视频的短回合及末段脱手一致。球2所需的连续200控制步窗口不可能在仅82控制步的交付回合内形成，故全程无达标区间、终态为false；此判断不以总分0或球体变色代替判据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.40–3.04 s，MyoHand单手的弯曲手指围住掌内的粉褐色球2，随手指屈曲使其在指掌间滚动、换位；视频3.12–3.28 s最终脱手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 3.12,
+   "log_step": 63,
+   "where": "操作中",
+   "evidence_zh": "观察：视频3.04 s时两球仍在弯曲手指附近，视频3.12–3.28 s黄色球1从指掌左侧迅速飞离，粉褐色球2也向另一侧脱离，结束前没有接回；两项跟踪并行，此处以球1明显脱手定位失败步骤。log step 63显示agent生成并保存的就是82步失败轨迹；评分器报告task_failure=1、sim_steps=82、deterministic=1。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 3.12,
+    "log_step": null,
+    "evidence_zh": "观察：视频0.80–3.04 s，MyoHand弯曲手指将两球留在指掌间并使其移动；视频3.12–3.28 s，左侧视角可见黄色球1迅速远离指掌，粉褐色球2也与手指拉开间隙，右侧近景可见球2下坠，均未重新承接。已有指掌接触后丢失承托，且不是到达目标后的计划松手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 3.12,
+    "log_step": 63,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 44及58的agent脚本以周期肌肉激励叠加两球位置误差反馈来驱动双球，目标与任务的跟踪要求一致；log step 61选择的参数仍只有64步共同连续合格，log step 63用该参数生成并保存的轨迹明确在82步失败。结合视频3.12–3.28 s手指仍弯曲而两球突然脱离支承，说明该肌肉控制未能持续维持球的稳定手内运动；现有证据只能归到控制，不能确定是接触布局、反馈增益还是其他执行细节。"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志；视频中的目标标记小且受球和手指遮挡，不能精确恢复两球三维误差、各自短暂合格窗口或掉球瞬间的世界高度。不过评分器的82步回合长度足以排除任何200步节点达成。评分器未报告target1_dist、target2_dist、solved_run，三项native_metrics均留null；日志中的64是agent脚本计算的共同最长窗口，不当作评分器原生指标。缺少接触力及控制对照试验，根因不细分。"
+ },
+ "summary_zh": "MyoHand已同时对两球进行手内调整，但交付回合仅82控制步，两项要求连续200步合格的跟踪节点均未达成，终态0/2。视频约3.12 s，球1迅速飞离指掌，球2也脱手，断点为球1持续跟踪的操作中，现象是没保持住，根因归为控制。缺少逐步状态和接触力，无法精确复核短暂跟踪误差或细分控制失稳原因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注与视频、日志一致：回合仅82控制步，两个200步节点均不可能达成，断点与现象（没保持住）有帧证据，根因控制已保守表述。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:58:27Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--y8ssqwd/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-privileged--y8ssqwd/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "8a2526dfb7b1f22bb1288ad741b225e54531d8f149ae5c39cfe91e5c6d5c2f10",
+   "video_sha256": "b5bcad7b5c6aa879cb36d9af3806a2593b223137da7b5b66994ff5a27eade25a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--4uq3ivr.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--4uq3ivr.json
new file mode 100644
index 0000000000000000000000000000000000000000..02097f4af1440cbcdac21c8ba38becb824f065d3
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--4uq3ivr.json
@@ -0,0 +1,261 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-standard--4uq3ivr",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 212,
+   "sim_steps": 212,
+   "physics_steps": 2120,
+   "steps": 229,
+   "calls": 4,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 229,
+   "elapsed_s": 2708.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.4,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.371,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-standard--4uq3ivr/replay.mp4",
+  "duration_s": 8.52,
+  "fps": 25.0,
+  "frames": 213,
+  "sha256": "59808fa95fef8b592e0a93655c6bd6c69fefc8db9d168522b093293e7fbdd1db"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.4,
+      "log_step": 37,
+      "text_zh": "观察：视频 0.96–3.36 s 的近景中，MyoHand转动腕掌并屈伸手指，两球仍在指掌间，但没有持续绕掌交换位置；log step 37 中 r14.log 的规划读数为控制步40球1误差0.0148 m、控制步60球1误差0.0248 m且run=0。这些是会话中的规划读数，不作为交付回合的实时状态；交付序列的球1失跟另由 log step 41 对 a14.npy 的诊断支持。"
+     },
+     {
+      "t": 5.6,
+      "log_step": 41,
+      "text_zh": "推断：视频 2.4–7.6 s 的球与手指相对运动，结合 log step 41 对最终交付 a14.npy 的诊断（索引39–69中球1绕掌角约82–91度，而目标角继续从115度到169度），支持球1未跟上移动目标。成功参考视频 2.4–6.8 s 则可见双球连续换位；该参考仅用于识别绕转表现。"
+     },
+     {
+      "t": 8.48,
+      "log_step": 46,
+      "text_zh": "推断：log step 46 明确将 a14.npy 重置后提交为第2回合，服务终止于212控制步；本视频含初态及212次推进，最后下落事件在视频 7.84–8.48 s，与该回合一致。一控制步对应一帧，即视频0.04 s，200步窗口跨7.96个视频秒；上述视频中段的跟踪偏离使整个短回合无法积满球1的200步窗口，故从未达到节点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.4–7.6 s，MyoHand单手的屈曲手指配合掌面持续接触掌内双球，通过转腕及手指开合尝试滚动、换位球1；没有换手或使用工具。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 5.6,
+      "log_step": 40,
+      "text_zh": "观察：视频 4.48–5.92 s 的近景中，一球长时间停在远端手指附近，另一球留在掌侧，没有像成功参考视频 4.4–6.4 s 那样继续绕掌换位。log step 40 是最终交付 a14.npy 的本地诊断：索引134–139中球2的归一化椭圆半径为2.46、2.36，而目标轨道半径为1；这些数值不是服务的实时球位姿。"
+     },
+     {
+      "t": 7.92,
+      "log_step": 46,
+      "text_zh": "观察：视频 7.76 s 时浅褐色球仍贴近伸出的手指，视频 7.84–8.08 s 它与手指间的间距迅速增大，视频 8.16–8.48 s 继续远离并下落，黄色球仍在掌指间；log step 46 的第2回合服务返回 failure: a ball dropped or left its target，control_steps=212。"
+     },
+     {
+      "t": 8.48,
+      "log_step": 46,
+      "text_zh": "推断：视频中段没有持续绕转，末段又失去一球，结合 log step 40 中交付文件的球2轨道偏离诊断，支持球2未形成连续200步合格窗口。log step 46 确认同一 a14.npy 的本地掉球和服务失败均在控制步212，已用视频 7.84–8.48 s 的掉球核对回合；212控制步仅跨8.48个视频秒，不能把短暂在掌内或变色当作持续200步达标。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.4–7.76 s，MyoHand单手手指配合掌面接触并滚动、承托双球，尝试在手内换位球2；视频约7.84 s后远端浅褐色球脱离指掌支承。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 2.0,
+   "log_step": 41,
+   "where": "操作中",
+   "evidence_zh": "推断：视频约1.6–2.4 s起，机器人继续转腕和屈指，但掌内两球没有随目标持续绕转，之后始终未修复为完整跟踪窗口；log step 41 对交付序列 a14.npy 的本地诊断显示，索引4–69中球1绕掌角约76–95度，而目标角从52度推进到169度，其中索引39–69球1仅约82–91度、目标角从115度推进到169度，支持球1未跟上移动目标。因此将最早未修复步骤定位为球1持续跟踪，视频约2.0 s为粗略锚点。log step 46 确认交付该文件，视频 7.84–8.48 s 后续掉球与服务212步失败一致；掉球不是另一个断点。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 2.4,
+    "log_step": 41,
+    "evidence_zh": "推断：视频 1.6–3.36 s 中MyoHand虽持续运动，但两球没有继续完成绕掌换位；视频 4.48–5.92 s 仍主要分居掌侧与远端手指。结合 log step 41 对已交付 a14.npy 的本地诊断，球1在索引39–69保持约82–91度的绕掌角，而其移动目标从115度转到169度，说明需要跟踪的运动目标已被落下；这是没跟上移动目标。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 16、34–35 的脚本明确以两球各自的移动目标误差搜索39个肌肉命令，并非瞄准了别的任务目标；log step 37 是 r14.log 的会话规划读数；log step 40–41 对最终交付 a14.npy 的本地诊断显示球1绕转停滞、球2随后偏出轨道。log step 46 把该序列提交给服务后同样在212控制步失败，视频 1.6–7.6 s 可见腕掌大幅转动却未实现持续双球换位，视频 7.84–8.48 s 又掉球。据此归到肌肉控制未实现所需滚动跟踪；现有材料不足以细分接触设计、轨迹求解或反馈控制。"
+   }
+  ],
+  "gap_zh": "没有逐控制步的实时球心、目标三维位置、接触对和各球连续合格计数；评分器也未报告 target1_dist、target2_dist 或 solved_run，因此原生指标均填null。小目标标记受球和手指遮挡，不能从变色精确确认0.015 m阈值；节点未达与球1先失跟的判断结合视频中的绕转缺失及最终交付文件的本地诊断作推断，不能把本地诊断等同于实时状态日志。球的固定编号与画面颜色对应也主要依赖该诊断的运动一致性，精确越界帧及控制子类无法独立核实。"
+ },
+ "summary_zh": "第2回合中，MyoHand通过转腕及手指开合在掌内操作双球，但两颗球均未完成连续200控制步跟踪，最终节点为0/2。球1先未跟上移动目标，之后视频约7.84–8.48 s一球脱离指掌支承并下落，服务在212控制步判失败，根因归为控制。没有实时三维球位姿或连续合格计数，精确距离、越界时刻和控制子类依赖交付文件的本地诊断辅助推断。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点0/2、断点球1、动态时机/控制整体有据可循（回合在212控制步掉球，视频7.84–8.48 s一致）。仅有一处证据归属不准：把r14.log的读数说成最终交付a14.npy的诊断。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点0/2、断点球1、动态时机/控制与视频及log一致；第1轮的证据归属问题已修正，未发现新的内容性错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:11:21Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--4uq3ivr/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--4uq3ivr/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--4uq3ivr/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--4uq3ivr/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "62f3e465a092bcb4a53f550a98009c0642032024498da1633acc3b2ca544183f",
+   "video_sha256": "59808fa95fef8b592e0a93655c6bd6c69fefc8db9d168522b093293e7fbdd1db",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--fvtkunk.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--fvtkunk.json
new file mode 100644
index 0000000000000000000000000000000000000000..de644afa30fb33523c0b7caa9095fe4c1905d117
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--fvtkunk.json
@@ -0,0 +1,240 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-standard--fvtkunk",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 319,
+   "sim_steps": 319,
+   "physics_steps": 3190,
+   "steps": 2104,
+   "calls": 2120,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 2104,
+   "elapsed_s": 3617.0,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.52,
+   "peak_processes": 9,
+   "peak_disk_gb": 1.251,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-standard--fvtkunk/replay.mp4",
+  "duration_s": 12.8,
+  "fps": 25.0,
+  "frames": 320,
+  "sha256": "e2383767340a7e07808528a878e59aa533e3954ff7ab00a904c614c502717213"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 91,
+      "text_zh": "观察：log step 91启动最后一次reset后的run_wave.py；log step 92的服务元数据确认episode=11、resets_left=0。视频0.0 s两球在掌中，随后MyoHand屈指、摆腕；本节点的200步持续条件在初始帧尚未累计完成。"
+     },
+     {
+      "t": 4.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.0–12.76 s的双视角及每约0.2 s近景显示，两球一直相邻留在指掌间，主要随手腕往复摆动，彼此相对方位只小幅改变；视频2.0、4.0、6.4、8.0、10.4 s未见球绕到另一侧的持续换位。参考成功视频0.0–8.28 s则可见两球沿掌内轨道交换方位。"
+     },
+     {
+      "t": 12.76,
+      "log_step": 93,
+      "text_zh": "推断：结合视频全程缺少完整绕转、视频11.8–12.76 s仍停留在相近掌内区域，以及任务中每个固定目标绕掌运行的要求，球1未形成连续200步的固定配对跟踪窗口，故intervals为空、final=false；不是仅由总成功分数为0推断。log step 93的agent自述“两球在振荡而未完成绕行”与画面一致，但未用其视觉估计充当精确距离。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.2–12.76 s，MyoHand单手屈曲的手指与掌面持续包托球1所在的双球组，随腕部摆动在手内小幅拨动、调整；球未脱离手掌，未完成持续绕转。不据遮挡画面细分每根手指的接触。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 91,
+      "text_zh": "观察：log step 91启动最后一次reset后的run_wave.py；log step 92的服务元数据确认episode=11、resets_left=0。视频0.0 s两球在掌中，随后MyoHand屈指、摆腕；本节点的200步持续条件在初始帧尚未累计完成。"
+     },
+     {
+      "t": 4.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.0–12.76 s的双视角及每约0.2 s近景显示，两球一直相邻留在指掌间，主要随手腕往复摆动，彼此相对方位只小幅改变；视频2.0、4.0、6.4、8.0、10.4 s未见球绕到另一侧的持续换位。参考成功视频0.0–8.28 s则可见两球沿掌内轨道交换方位。"
+     },
+     {
+      "t": 12.76,
+      "log_step": 93,
+      "text_zh": "推断：结合视频全程缺少完整绕转、视频11.8–12.76 s仍停留在相近掌内区域，以及任务中每个固定目标绕掌运行的要求，球2未形成连续200步的固定配对跟踪窗口，故intervals为空、final=false；不是仅由总成功分数为0推断。log step 93的agent自述“两球在振荡而未完成绕行”与画面一致，但未用其视觉估计充当精确距离。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.2–12.76 s，MyoHand单手屈曲的手指与掌面持续包托球2所在的双球组，随腕部摆动在手内小幅拨动、调整；球未脱离手掌，未完成持续绕转。不据遮挡画面细分每根手指的接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 2.0,
+   "log_step": 91,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约2.0 s手已包住两球并摆腕，但双球仍保持相邻的近似方位；随后至视频12.76 s仅反复小幅摆动，未恢复为参考成功视频中的掌内绕转。两节点并行同时受阻，以球1持续跟踪作为断点；没有证据显示球2先于球1发生独立失败。log step 91为该评分回合的启动记录，前一回合在log step 90的掉球不属于本次视频。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 2.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频约2.0–12.76 s，MyoHand手腕和手指持续活动，两球却只在掌中往复小幅移动，未随需要绕掌运动的目标持续换位；视频11.8–12.76 s仍如此。推断：这是对移动目标未跟上的表现；未见本回合掉球，不将调试回合掉球标为没保持住。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 91,
+    "source_zh": "日志",
+    "evidence_zh": "观察：log step 86的run_wave.py用control_steps生成固定正弦相位和手腕、手指关节目标，再以本体感知计算肌肉命令；BallTracker的球位置只在每20步的分支内检测、打印及保存，不进入下一步动作计算。log step 89、91仅修改固定摆幅和屈指姿态，并在log step 91启动最终评分回合；log step 93的agent自述仍为振荡。推断：该关节周期动作没有根据实际球与移动目标的偏差修正掌内运动，视频全程只有振荡而未形成绕转，说明所实现控制未能产生所需双球运动。现有证据不足以进一步区分接触几何、轨迹或底层肌肉跟踪问题，因此停在控制。"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志，也没有评分器报告的target1_dist、target2_dist或solved_run；小目标标记与部分指球接触被遮挡，不能从二维画面或变色复原三维0.015米误差、短暂合格时段及精确首次越界时刻。两球均未形成可见的绕转窗口，故可判两个持续节点未完成；断点视频约2.0 s仅为振荡停滞的导航锚点，无法判两节点失败的先后，也不能将画面颜色可靠地绑定object编号。",
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  }
+ },
+ "summary_zh": "评分采用log step 91最后重置后的第11回合：MyoHand持续包托并小幅调整双球，但视频0.0–12.76 s均未形成跟随移动目标的完整绕转，两个持续跟踪节点均未达到，最终0/2。断点记球1持续跟踪的操作中，球2并行受阻；固定周期关节控制只产生振荡，球位置检测未用于动作纠偏，评分器最终记录timeout=1、live_success=0。缺少逐步三维距离与合格计数，无法量化瞬时误差或精确定位第一次跟踪越界。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "未发现需要修改的错误：两节点final=false与视频一致（两球全程留在掌中小幅摆动），断点、现象与根因证据充分且标注了推断，缺口说明合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:59:17Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--fvtkunk/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--fvtkunk/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1f863a7f5a84873642684abaf23247e749fd17c71e0be23c2ecb19ab48fc044e",
+   "video_sha256": "e2383767340a7e07808528a878e59aa533e3954ff7ab00a904c614c502717213",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--ghmmw85.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--ghmmw85.json
new file mode 100644
index 0000000000000000000000000000000000000000..37569ede22c3e88a7383c89011d3d77c17850463
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--ghmmw85.json
@@ -0,0 +1,214 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-standard--ghmmw85",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 9,
+   "sim_steps": 9,
+   "physics_steps": 90,
+   "steps": 145,
+   "calls": 134,
+   "episodes": 7,
+   "resets": 6,
+   "reset_budget": 10,
+   "sim_steps_total": 145,
+   "elapsed_s": 3489.8,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.995,
+   "peak_processes": 12,
+   "peak_disk_gb": 0.383,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-standard--ghmmw85/replay.mp4",
+  "duration_s": 0.4,
+  "fps": 25.0,
+  "frames": 10,
+  "sha256": "2b6cd4c4ed618920c0e572a7a98b3a19bf1246ff09847accc4571c9d261b2924"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00–0.36 s的全部10帧中，MyoHand由较伸展姿态逐渐屈指成杯状，两颗球始终位于指掌附近，球1没有出现持续绕行一整圈的过程。"
+     },
+     {
+      "t": 0.36,
+      "log_step": null,
+      "text_zh": "推断：评分器reward仅记录最后回合n_actions=9、sim_steps=9，视频恰为初始画面加9步的10帧；即使球1每步都满足距离阈值，也不可能累计节点所需的200个连续控制步，故全程未达到。视频时间按25 fps读取，未将agent工作时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.00–0.36 s，MyoHand单只手的手指在球1周围屈曲，与掌面配合承托、挪动该球；两球仍在指掌间，没有可见的主动释放。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00–0.36 s的全部10帧中，MyoHand由较伸展姿态逐渐屈指成杯状，两颗球始终位于指掌附近，球2没有出现持续绕行一整圈的过程。"
+     },
+     {
+      "t": 0.36,
+      "log_step": null,
+      "text_zh": "推断：评分器reward仅记录最后回合n_actions=9、sim_steps=9，视频恰为初始画面加9步的10帧；即使球2每步都满足距离阈值，也不可能累计节点所需的200个连续控制步，故全程未达到。视频时间按25 fps读取，未将agent工作时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.00–0.36 s，MyoHand单只手的手指在球2周围屈曲，与掌面配合承托、挪动该球；两球仍在指掌间，没有可见的主动释放。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": null,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频0.00–0.36 s，MyoHand已在两球周围屈指并进行掌内操作；评分器记录最后回合只有9个控制步、success=0、timeout=1、task_failure=0。推断：球1持续跟踪尚未完成且此后没有修复，断点定位该操作；球2的并行跟踪同样未完成。录像没有呈现可单独定位的不可逆物理失败时刻，因此时间置空。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "观察：inputs/含本次录像、完整会话日志和成功参考录像，但没有states.jsonl；评分器没有报告target1_dist、target2_dist或solved_run，原生指标均置null。本次视频0.00–0.36 s仅展示9个控制步，末帧仍见两球处于指掌附近；低分辨率及遮挡不能确定精确三维跟踪误差，也未见明确掉球或其他可归类的失败现象，故现象层留空。log step 64、66含先前回合的服务终止信息，log step 67为agent对会话的总结，均不能证明最后9步回合发生同样故障；缺少最后回合的逐步估计与执行误差，无法区分感知、接触设计和反馈控制等根因，故根因层留空。",
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  }
+ },
+ "summary_zh": "评分录像中MyoHand已对掌中两球进行屈指与手内调整，但最后回合只有9个控制步，两个要求连续200步跟踪的节点均未达到，终态为0/2。断点定位球1持续跟踪的操作中，球2并行操作也未完成；视频0.36 s末帧仍见两球处于指掌附近，不能把先前测试回合的掉球套用于本次评分轨迹。缺少最后回合的三维球位、目标误差和连续合格计数，具体失败现象与根因留空。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注与证据一致：评分回合只有9个控制步（n_actions=9，视频10帧0.4 s），两个要求连续200步的节点不可能达到，final=0；断点、空现象/根因层均带gap_zh，且未把先前回合的掉球套用到评分回合。未发现需要改动的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:20Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--ghmmw85/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--ghmmw85/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "8aa3281885d290ea255e22159d52cb8c75174e63cbe0dc34596b991773ff874d",
+   "video_sha256": "2b6cd4c4ed618920c0e572a7a98b3a19bf1246ff09847accc4571c9d261b2924",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--jisq83e.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--jisq83e.json
new file mode 100644
index 0000000000000000000000000000000000000000..7469f70d9662f30bc6e5ede740d0701dc150d0e9
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--jisq83e.json
@@ -0,0 +1,257 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-standard--jisq83e",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 271,
+   "sim_steps": 271,
+   "physics_steps": 2710,
+   "steps": 341,
+   "calls": 15,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 341,
+   "elapsed_s": 1937.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.526,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.663,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-standard--jisq83e/replay.mp4",
+  "duration_s": 10.88,
+  "fps": 25.0,
+  "frames": 272,
+  "sha256": "2ef4693c093a60ac620f3439a80e6c096a8d9eeec04c6ae1a27b93060b921ddf"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.4,
+      "log_step": null,
+      "text_zh": "观察：视频0.0–10.84 s的双视角中，单手屈指托住黄色球1和粉色球2，但两球长时间保持同侧相邻位置；视频6.0–7.2 s右侧近景可见黄色目标标记经过粉色球附近，黄色球1仍在其右侧，未随对应标记换位。成功参考视频3.0–5.0 s则可见两球随目标交换掌中位置。"
+     },
+     {
+      "t": 10.84,
+      "log_step": 318,
+      "text_zh": "观察：视频10.52–10.84 s，黄色球1脱离指掌，左侧全景可见球与手之间的间距迅速增大；log step 318中，机器人服务在最后第3回合control_steps=271报告done=True、reason=failure: a ball dropped or left its target。"
+     },
+     {
+      "t": 6.4,
+      "log_step": 318,
+      "text_zh": "推断：log step 318的271个控制步对应录像初态加271帧（272帧、25 fps），末步271/25=视频10.84 s与可见脱手终止核对一致；200个连续步跨199/25=7.96个视频秒。视频6.4 s球1明显偏离配对目标，任何可能的完整200步窗口都包含此段偏离，因此没有完成保持节点；这里未用变色或零分代替位置判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.4–10.4 s，MyoHand单手的屈曲手指配合掌面持续接触黄色球1，随腕姿变化使其在指掌间滚动；视频10.52–10.84 s球1脱手。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.4,
+      "log_step": null,
+      "text_zh": "观察：视频0.4–7.2 s，粉色球2持续位于黄色球1左侧的指掌处，未随移动目标绕过另一颗球；视频6.4 s右侧近景中，两球及小目标标记的相对位置已错开，黄色小标记位于粉色球附近，另一配对位置在黄色球一侧。成功参考视频3.0–5.0 s呈现两球交换位置，本次没有该绕转过程。"
+     },
+     {
+      "t": 10.84,
+      "log_step": 318,
+      "text_zh": "推断：视频6.0–7.2 s的配对目标与球2错位排除了连续200步跟踪；该偏离段位于本录像所有可能的7.96视频秒保持窗口内（log step 318的271控制步与272帧录像末端脱手相符，每控制步一帧、25 fps）。视频10.84 s粉色球2仍在手边，但留在掌中不足以满足移动目标跟踪；未把本地测试数值当作本回合逐步测量。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.4–10.84 s，MyoHand单手的弯曲手指和掌面包围、承托粉色球2，随手腕转动在手内小幅滚动换位；没有双手协作或工具接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 3.2,
+   "log_step": 318,
+   "where": "操作中",
+   "evidence_zh": "推断：视频2.4–4.0 s机器人已在屈指、转腕操作双球，但两球相对指掌的位置变化很小，未继续沿移动目标绕行；视频6.0–7.2 s可见配对目标与球明显错位，此后也未形成完整跟踪窗口。两个跟踪步骤并行，取黄色球1的持续跟踪作为断点，视频3.2 s仅为较早停滞段的导航锚点，并非精确越过0.015 m的时刻；视频10.52–10.84 s球1最终脱手，log step 318的第3回合在271控制步终止。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 3.2,
+    "log_step": null,
+    "evidence_zh": "观察：视频2.4–7.2 s机器人虽持续屈指和转腕，两球仍大体维持原来的左右关系，移动目标已绕向另一侧；视频6.4 s黄色小目标已到粉色球附近，黄色球1没有跟上。目标在移动，双球未持续同步绕行。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 282,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 282交付脚本按固定200步周期的基频及二次谐波预生成39维肌肉动作，执行时每50步提交，仅检查服务done/reason，没有按球与目标位置修正动作；log step 318确认第3回合执行该脚本及其谐波参数。该肌肉控制轨迹未使双球持续绕转，并在视频10.52–10.84 s失去对球1的保持；log step 315的独立本地测试也只得到最长48步共同合格、270步存活，作为控制能力的辅助证据，不能作为本次视频的精确计数。现有材料不足以进一步区分接触布局、轨迹或具体肌肉参数问题。"
+   }
+  ],
+  "gap_zh": "观察：inputs/中有本次录像、运行日志和成功参考录像，但没有逐控制步状态日志；评分器reward未报告target1_dist、target2_dist或solved_run，故三个原生指标均为null。推断：视频足以排除完整跟踪窗口并确认黄色球1脱手，但遮挡和二维投影不能给出精确三维误差、最早越界帧及两个并行节点谁先越界；断点时间是可见停滞段的粗略锚点。log step 315等本地仿真结果仅用于根因辅助，不与评分回合合并。"
+ },
+ "summary_zh": "观察：最后第3回合中，MyoHand单手接触并调整双球，但两球没有持续随各自移动目标绕行，两个200步跟踪节点均未完成，终态0/2。推断：断点取球1持续跟踪，固定谐波肌肉控制未维持同步绕转；视频10.52–10.84 s黄色球1脱手，log step 318的机器人服务在271控制步报告失败。观察：缺少本回合逐步三维距离与合格计数，无法确定精确越界时刻及更细控制根因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点值（两个跟踪节点未达成、final=0）与断点选择可接受；但把评分器终止事件（球脱手）单列为第二个现象不合规则，应删除，改为在证据中提及。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点值（两个跟踪节点未达成、final=0）、断点（球1持续跟踪，操作中）、现象与根因均可接受；第1轮意见已处理，时钟换算（271控制步=272帧）与末帧脱手一致。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:10:51Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--jisq83e/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--jisq83e/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--jisq83e/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--jisq83e/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3c2aaf16c9c305bb88b54f25af00f513817afb35eb9a3f13a0221658fcc4eea0",
+   "video_sha256": "2ef4693c093a60ac620f3439a80e6c096a8d9eeec04c6ae1a27b93060b921ddf",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--pntstup.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--pntstup.json
new file mode 100644
index 0000000000000000000000000000000000000000..bc1f23cba16fad0895b8b79255f52330e1f0839f
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--pntstup.json
@@ -0,0 +1,229 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-standard--pntstup",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 60,
+   "sim_steps": 60,
+   "physics_steps": 600,
+   "steps": 96,
+   "calls": 5,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 96,
+   "elapsed_s": 1450.5,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 4.283,
+   "peak_processes": 20,
+   "peak_disk_gb": 1.322,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-standard--pntstup/replay.mp4",
+  "duration_s": 2.44,
+  "fps": 25.0,
+  "frames": 61,
+  "sha256": "ff223d8c921ea2ef5260ebe0f3acbc1d4cc2da2a3d6a9ff91b61dcdd6becd3e7"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 106,
+      "text_zh": "推断：log step 106 的服务返回确认评分对象为重置后的 episode 2，共执行60个控制步即失败；60步不足节点要求的连续200步，因此该节点全程未达到，不能把掌中短时靠近目标算作完成。视频0.00–2.40 s为该回合起止帧，已核对末段粉色球离手；61帧对应初始帧加60步，每步占视频0.04 s（60/25=视频2.40 s），不是每步0.025 s的仿真时钟。"
+     },
+     {
+      "t": 2.4,
+      "log_step": null,
+      "text_zh": "观察：视频0.00、0.40、0.80、1.20、1.60、2.00及2.40 s，黄色球随MyoHand指掌姿态改变而在手中移动，末帧仍在指掌围成的区域；这只能说明有操球，不能证明连续200步的三维目标跟踪。log step 38 的模型片段将ball1对应MatBallY。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.00–2.40 s，MyoHand单手的手指弯曲并配合掌面改变黄色球在手中的位置；球一直留在指掌区域，未见主动释放。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 106,
+      "text_zh": "推断：log step 106 的服务返回确认评分对象为重置后的 episode 2，共执行60个控制步即失败；60步不足节点要求的连续200步，因此该节点全程未达到，不能把掌中短时靠近目标算作完成。视频0.00–2.40 s为该回合起止帧，已核对末段粉色球离手；61帧对应初始帧加60步，每步占视频0.04 s（60/25=视频2.40 s），不是每步0.025 s的仿真时钟。"
+     },
+     {
+      "t": 2.16,
+      "log_step": 38,
+      "text_zh": "观察：log step 38 的模型片段将ball2对应MatBallP；视频0.00–2.08 s粉色球由指掌承托，视频2.12–2.24 s逐渐脱离指掌区域，视频2.40 s左侧全景中已远离手部，未被接回。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.00–2.08 s，MyoHand单手弯曲手指并配合掌面移动粉色球；视频约2.12–2.24 s球从指掌区域逸出，此后失去承托。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball2_target_tracked",
+   "t": 2.16,
+   "log_step": 106,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约2.12–2.24 s，MyoHand正在掌中操动双球时，粉色球离开指掌区域；视频2.40 s左侧全景显示它已远离手，黄色球仍在掌中。log step 106 的服务返回episode 2在60个控制步结束，reason为“failure: a ball dropped or left its target”。推断：两球跟踪为并行操作，此处以发生不可恢复失球的球2持续跟踪节点作为断点。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.16,
+    "log_step": 106,
+    "evidence_zh": "观察：视频1.76–2.08 s粉色球仍靠在MyoHand指掌间，视频2.12–2.24 s与指掌的距离迅速增大，视频2.40 s左侧全景可见其已脱离手部；机器人没有接回。log step 106服务同时报告该评分回合以掉球或离开目标失败结束。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "观察：log step 104中agent脚本生成并保存最终肌肉命令序列，本地rollout输出surv=60、nsucc=0、maxrun=0；这些是会话测试结果。log step 106中部署脚本重置后向服务开环播放同一序列，服务在60个控制步报失败；视频约2.12–2.40 s实际显示粉色球逸出指掌。推断：所生成的肌肉控制未能维持操球所需的承托与运动稳定性；本地测试与实时评分均暴露失球，支持控制层归因。没有足够证据区分接触设计、反馈律或参数问题，保留在控制总类。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "无逐控制步状态日志，服务未报告两球各自的目标距离、世界高度、共同连续合格计数或接触力；native_metrics三项均无评分器数值，故填null。二维画面不能精确定位0.015米越界或1.25米高度阈值的时刻，但评分回合只有60步，足以确定两个200步节点均未达到。根因仅能确定至控制总类，不能验证具体控制子类。"
+ },
+ "summary_zh": "评分对象为重置后的第2回合，MyoHand确实在掌中操动双球，但仅执行60个控制步，两个要求连续200步的跟踪节点均未达到，最终0/2。视频约2.12–2.24 s粉色球脱离指掌且未被接回，断点为球2持续跟踪，现象为没保持住；最终序列的本地测试和服务执行均在60步失效，支持控制层归因。缺少逐步距离、高度及接触数据，无法细分控制根因；成功参考视频仅用于比较视角与操球外观。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "评分回合为重置后的第2回合（log step 108状态：episode 2，control_steps 60，failure）。两节点未达到、断点球2、现象没保持住（粉球约2.12–2.24 s脱手）与视频一致；根因停在控制总类有日志支持（本地孪生与服务均在60步失败），未发现需改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:59:14Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--pntstup/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--pntstup/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f66116c4cef7f06858652953a067d5cbb9c73d8d0154026238d8f961b18018de",
+   "video_sha256": "ff223d8c921ea2ef5260ebe0f3acbc1d4cc2da2a3d6a9ff91b61dcdd6becd3e7",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--q9ytca9.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--q9ytca9.json
new file mode 100644
index 0000000000000000000000000000000000000000..95331749a272796a039ada50547a10b27b2cbeeb
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--q9ytca9.json
@@ -0,0 +1,262 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-standard--q9ytca9",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 70,
+   "sim_steps": 70,
+   "physics_steps": 700,
+   "steps": 778,
+   "calls": 98,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 778,
+   "elapsed_s": 1752.7,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.477,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.857,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-standard--q9ytca9/replay.mp4",
+  "duration_s": 2.84,
+  "fps": 25.0,
+  "frames": 71,
+  "sha256": "3e5d90c182fa94210ffe5e7c1e73f31794db3c4f19a9400a4993a7fd125af624"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 147,
+      "text_zh": "观察：log step 147 的服务回复显示最后一次重置进入 episode 11、resets_left=0；log step 149 执行控制脚本，评分视频从双球在指掌间的初态开始。此前回合的调试不计入本节点。"
+     },
+     {
+      "t": 2.8,
+      "log_step": 149,
+      "text_zh": "观察：视频 0.00–2.56 s，MyoHand 单手指掌小幅运动，两球仍在指掌附近；视频 2.60–2.80 s 粉色球逐渐离开支承，末帧黄色球仍在手边。评分器 reward.sim_steps=70、n_actions=70，live_matches_replay=1；视频共71帧。"
+     },
+     {
+      "t": 2.8,
+      "log_step": null,
+      "text_zh": "推断：本评分回合仅执行70个控制步，即使本球每一步都满足距离阈值，也不足节点要求的连续200步，所以整个视频内均未达到持续跟踪条件，终值为false；并非仅由success=0推定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频 0.00–2.40 s，MyoHand 单只手的指掌承托本节点对应球，随腕部及手指的小幅运动使球在手边滚动；没有清楚的合手夹紧动作。两球均受到承托，颜色与object编号未在输入中明确对应。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 147,
+      "text_zh": "观察：log step 147 的服务回复显示最后一次重置进入 episode 11、resets_left=0；log step 149 执行控制脚本，评分视频从双球在指掌间的初态开始。此前回合的调试不计入本节点。"
+     },
+     {
+      "t": 2.8,
+      "log_step": 149,
+      "text_zh": "观察：视频 0.00–2.56 s，MyoHand 单手指掌小幅运动，两球仍在指掌附近；视频 2.60–2.80 s 粉色球逐渐离开支承，末帧黄色球仍在手边。评分器 reward.sim_steps=70、n_actions=70，live_matches_replay=1；视频共71帧。"
+     },
+     {
+      "t": 2.8,
+      "log_step": null,
+      "text_zh": "推断：本评分回合仅执行70个控制步，即使本球每一步都满足距离阈值，也不足节点要求的连续200步，所以整个视频内均未达到持续跟踪条件，终值为false；并非仅由success=0推定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频 0.00–2.40 s，MyoHand 单只手的指掌承托本节点对应球，随腕部及手指的小幅运动使球在手边滚动；没有清楚的合手夹紧动作。两球均受到承托，颜色与object编号未在输入中明确对应。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 2.6,
+   "log_step": 149,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 2.40–2.56 s 粉色球仍贴近指掌边缘，视频 2.60–2.80 s 它与指掌的间隙迅速扩大并离开右侧画面，机器人未接回；黄色球仍在手边。评分器 reward.sim_steps=70、n_actions=70，少于两个节点各自要求的连续200步。推断：两个并行跟踪节点在同一段双球操作中均未完成，未见可区分的失败先后，按DAG所列第一个节点ball1_target_tracked记录断点，球2节点同样未完成；此记录不表示粉色球就是object1，颜色与object编号仍无明确对应，也无法确认两球各自首次误差越界的时刻。"
+  },
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.6,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–2.40 s 粉色球受MyoHand指掌承托；视频 2.60–2.80 s 粉色球脱离指掌并迅速远离，机器人没有恢复接触。此为原有承托丢失，未见主动将球放到目标上的释放。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 2.6,
+    "log_step": 146,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 146 的控制脚本读取相机与本体感知，以每次5个控制步的肌肉脉冲驱动手部；log step 149 记录实际执行14次submit。视频 0.00–2.56 s 指掌虽有运动，却没有维持住双球的绕转与支承，视频 2.60–2.80 s 粉色球滑脱，支持肌肉控制未能维持操作的判断；没有充分证据细分到接触设计、反馈控制或具体参数。"
+   }
+  ],
+  "gap_zh": "观察：inputs/提供本次和成功参照视频、完整会话日志，但没有状态日志；评分器未报告target1_dist、target2_dist或solved_run。log step 146 的脚本中间有截断，log step 149–150未收录最终控制器的打印输出；log step 151仅有agent自述“粉色球低于1.25米”，不能当作精确高度测量。推断：两节点可由70步不足200步确定未完成，但无法复核每步三维误差、精确掉落高度及控制子类；颜色与object编号无明确对应，不能把粉色球脱离支承归到某一个物体编号。断点按两个并行未完成节点中的首个节点记录，不表示已识别掉落球编号或失败先后。"
+ },
+ "summary_zh": "本次评分对象是最后的episode 11：机器人单手承托并尝试操纵双球，但仅运行70个控制步，两个要求连续200步的持续跟踪节点均未完成，终态0/2。视频约2.60–2.80 s粉色球脱离指掌且未接回，现象为没保持住，证据支持一般控制问题；两个节点并行未完成，断点按DAG首个节点ball1_target_tracked记录。没有逐步误差与高度日志，且颜色与object编号未明确对应，不能确定掉落球编号、两球首次越界先后或控制子类。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点终值false、时钟（70控制步≈71帧）、粉球约2.6 s脱离指掌均与视频一致；唯一问题是断点ref为unknown，而两个并行节点同时失败，应取其一。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "两节点终值false（70控制步<200步）、粉球约2.6 s脱离指掌、断点ref取首个并行节点、根因控制均与视频和评分字段一致，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:25Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--q9ytca9/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--q9ytca9/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--q9ytca9/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--q9ytca9/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "5fc5fa4ab6bf32ab075fe28a7712aeb6bd7a017e65a01565a03b70bb11f1faee",
+   "video_sha256": "3e5d90c182fa94210ffe5e7c1e73f31794db3c4f19a9400a4993a7fd125af624",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--qftdcfj.json b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--qftdcfj.json
new file mode 100644
index 0000000000000000000000000000000000000000..cb605266c1c0305dec82a6b1fc3aaf3360e8235a
--- /dev/null
+++ b/phase2/myosuite/baoding/trajectories/myosuite-baoding-i00-standard--qftdcfj.json
@@ -0,0 +1,230 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-baoding-i00-standard--qftdcfj",
+ "taskKey": "myosuite/baoding",
+ "dag_sha256": "d601c830f12d5f8aa686b3f3d6b6e58345f56dd2da1103766529b1f1c3ddafd8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 1000,
+   "sim_steps": 1000,
+   "physics_steps": 10000,
+   "steps": 1036,
+   "calls": 6,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 1036,
+   "elapsed_s": 2381.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.721,
+   "peak_processes": 15,
+   "peak_disk_gb": 0.501,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-baoding-i00-standard--qftdcfj/replay.mp4",
+  "duration_s": 40.04,
+  "fps": 25.0,
+  "frames": 1001,
+  "sha256": "34cf003498971fba7fa3eaa90c500baa22ec9beb7d83de78149ad93fc988c906"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball1_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.0、0.4、1.2、2.4、4.0、6.4、8.0 s的两侧视角显示，MyoHand先收拢手指并使双球小幅移位，随后两球分别停在指掌间的固定区域；小目标标记继续绕行，球没有随之持续换位。球1对应的持续跟踪条件未见达成，球体变色不作为达标依据。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 69,
+      "text_zh": "推断：视频9.6–38.4 s每约0.8 s的近景以及视频39.04–40.00 s的密集末帧仍显示双球维持原有相对位置，没有任何完整目标周期的绕行；因此球1未连续200个控制步跟随其固定配对目标，最终为false。log step 69的机器人服务确认评分第2回合执行至control_steps=1000，因目标轨迹耗尽结束；这支持未成功，但两球各自未完成的判断来自上述画面，非仅由零分反推。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频0.4–4.0 s及视频39.04–40.00 s，单只MyoHand弯曲的手指配合掌面包夹、留住球1；起始有小幅位移，随后主要保持该握持位置，未见持续滚动换位。"
+     }
+    ]
+   },
+   "ball2_target_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.0、0.4、1.2、2.4、4.0、6.4、8.0 s的两侧视角显示，MyoHand先收拢手指并使双球小幅移位，随后两球分别停在指掌间的固定区域；小目标标记继续绕行，球没有随之持续换位。球2对应的持续跟踪条件未见达成，球体变色不作为达标依据。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 69,
+      "text_zh": "推断：视频9.6–38.4 s每约0.8 s的近景以及视频39.04–40.00 s的密集末帧仍显示双球维持原有相对位置，没有任何完整目标周期的绕行；因此球2未连续200个控制步跟随其固定配对目标，最终为false。log step 69的机器人服务确认评分第2回合执行至control_steps=1000，因目标轨迹耗尽结束；这支持未成功，但两球各自未完成的判断来自上述画面，非仅由零分反推。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频0.4–4.0 s及视频39.04–40.00 s，单只MyoHand弯曲的手指配合掌面包夹、留住球2；起始有小幅位移，随后主要保持该握持位置，未见持续滚动换位。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "target1_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "target2_dist",
+   "unit": "m",
+   "threshold": 0.015,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 200,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball1_target_tracked",
+   "t": 2.4,
+   "log_step": 69,
+   "where": "操作中",
+   "evidence_zh": "观察：视频0.4–2.4 s手指收拢后双球停在指掌间，视频2.4–9.2 s目标标记移动而球未完成相应绕行，后续直至视频40.00 s也未修复。两个并行跟踪步骤遇到同一问题，以球1持续跟踪作为断点；log step 69表明这正是服务重置后的评分第2回合，结束原因为1000个控制步的目标轨迹耗尽，未发生视频中可见的掉球。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 2.4,
+    "log_step": null,
+    "evidence_zh": "观察：视频0.4–9.2 s的近景中，MyoHand使双球稳定在两个近乎固定的位置，小目标标记从球旁及球面投影处经过；视频9.6–38.4 s仍反复出现相同情况。球没有跟上运动中的目标，不是失去持球接触；视频39.04–40.00 s两球仍留在手中。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "log_step": 69,
+    "evidence_zh": "推断：log step 38、43、58中的脚本把静态握持肌肉命令叠加周期性傅里叶项，按预先生成的命令序列执行；log step 69实际向机器人服务提交该B2策略的1000行命令，首段400个控制步的手部关节轨迹与离线测试比较最大差为0.00000。视频0.4–9.2 s及视频9.6–38.4 s显示策略只产生小幅手内运动，未形成双球轮流绕转，说明所找到的肌肉控制序列不足以实现滚动跟踪。log step 53、68、85的离线诊断及变体测试也显示握持附近的控制未能得到持续绕行，但这些测试不作为评分回合的节点值或掉球现象。",
+    "note_zh": "控制序列未能产生所需绕转有直接证据；没有接触力及逐步球状态，无法进一步区分抓取接触布局、肌肉激活幅相和反馈控制的贡献，保留上位控制标签。"
+   }
+  ],
+  "gap_zh": "本输入没有状态日志；两侧视频可判明未持续绕转，但小目标部分被手指和球遮挡，不能测得各球逐控制步的三维距离、独立连续合格计数或精确越界时刻。log step 69的final dists和best solved streak=24来自agent脚本的离线模拟，不是机器人服务或评分器报告的距离与计数；不据此填写原生指标，也不据此设定节点区间。两球并行出现同一失败，无法确定谁先越过数值阈值，断点时间仅锚定早期可见的未跟随。",
+  "unmet": {
+   "ball1_target_tracked": false,
+   "ball2_target_tracked": false
+  }
+ },
+ "summary_zh": "评分第2回合中，MyoHand握住双球并执行满1000个控制步，但双球主要停在掌内固定位置，没有持续跟随绕行目标，两个节点均未完成。断点取并行的球1持续跟踪，现象为未跟上移动目标，日志与画面支持肌肉控制序列未能形成绕转，服务最终因目标轨迹耗尽结束。没有逐步球状态和评分器距离值，精确误差、连续合格时长及控制子类无法核实。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "两球节点均未达成、断点与动态时机现象、控制根因与帧和服务读数一致，无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:02:36Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--qftdcfj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-baoding-i00-standard--qftdcfj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c83e3e951e8d7ba48f87e6a231f4b88e3301c45163e8b515d24fd892c2b0200c",
+   "video_sha256": "34cf003498971fba7fa3eaa90c500baa22ec9beb7d83de78149ad93fc988c906",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--ahkgq8f.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--ahkgq8f.json
new file mode 100644
index 0000000000000000000000000000000000000000..32550a82f0c719906a676f0c6a5eb248d04e6224
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--ahkgq8f.json
@@ -0,0 +1,240 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-privileged--ahkgq8f",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 471,
+   "sim_steps": 471,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 532,
+   "agent_resets": 575,
+   "agent_sim_steps": 338500,
+   "agent_records": 463,
+   "agent_replays": 1,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 6.302,
+   "peak_processes": 39,
+   "peak_disk_gb": 0.395,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-privileged--ahkgq8f/replay.mp4",
+  "duration_s": 18.88,
+  "fps": 25.0,
+  "frames": 472,
+  "sha256": "881dbc4fcab865b390e5376a7bc1f95ac6cb545288763daa048c4983d834c6e7"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [
+     [
+      12.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 8.0,
+      "log_step": null,
+      "text_zh": "观察：视频 8.0–12.6 s 的俯视画面中，肌肉右手的拇指与其余手指从相对两侧夹住红边灰色长方体；视频 13.2–15.8 s 物体离开起始柱并随右手抬升，形成实际抓握。"
+     },
+     {
+      "t": 12.5,
+      "log_step": 106,
+      "text_zh": "推断：log step 106 交付并回放的文件有471个动作，本视频为472帧、25 fps，含初态帧，故100个控制步对应约4 s视频；结合视频约 8.0 s 开始的夹持接触及视频 13.2–15.8 s 的持物抬升，右臂累计接触100步约在视频 12.5 s 达标（估计误差约0.5 s，无逐步接触计数）。该节点按DAG是历史事件，视频 16.4 s 滑脱不撤销已达记录。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 8.0–16.2 s，肌肉右手的拇指与其余手指夹住红边长方体，随后将其抬离起始柱；视频 16.4 s 物体滑出，白色假肢手未参与。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 16.4,
+      "log_step": null,
+      "text_zh": "观察：本次视频 0–18.84 s 只有肌肉右手围绕起始柱抓取和抬升；视频 16.2–16.8 s 物体从右手滑出，未由白色假肢手承接，视频 17.0–18.84 s 也未见补接。成功参照视频 16.0–19.0 s 可见白手伸到物体处承接，本次没有出现这一动作或物体随白手移动的结果。"
+     },
+     {
+      "t": 18.84,
+      "log_step": 106,
+      "text_zh": "观察：log step 106 的交付生成器只执行肌肉手抓取抬升脚本，其实际回放返回 success=False、done=True、steps=471；该结果与视频末段掉落一致。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 18.84,
+      "log_step": null,
+      "text_zh": "观察：视频 0–18.84 s 物体始终在起始柱及肌肉右手附近，视频 16.4–18.84 s 脱手后向柱侧下方掉落，没有被左手运到另一根目标柱顶，也没有松手留在目标柱上的阶段。成功参照视频 24.0–31.12 s 可见物体留在另一柱顶、假肢手离开的结果，本次未出现。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_held_by_myo",
+   "t": 16.3,
+   "log_step": 106,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 15.8–16.2 s 肌肉右手仍夹着离柱的长方体，视频 16.4 s 手与物体分开，视频 16.8–18.84 s 物体继续翻转下落，未被重新抓住或由白手接住。log step 106 对交付文件的回放在471个控制步结束，返回 failure: object dropped or the time limit passed。推断：机器人在右手抓取抬升阶段未保持住物体，断点落在该步；累计接触历史达标不等于抓稳或已开始交接。"
+  },
+  "unmet": {
+   "object_held_by_myo": true,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 16.3,
+    "log_step": null,
+    "evidence_zh": "观察：视频 13.2–15.8 s 长方体被肌肉右手夹持并抬离起始柱；视频 16.2–16.4 s 它从弯曲的手指间滑出，随后翻转下落，到视频 18.84 s 仍未被承接。此前已有实际持物接触，失败发生在保持抓握阶段。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频 13.2–16.4 s 显示肌肉右手能到达物体、闭合并抬起它，但物体在指间转动并滑出，表明执行中的抓握控制未能保持承载接触。日志中的交付生成器（log step 106）确实使用抓取抬升脚本，agent 也在 log step 107 自述手指和拇指跟踪欠佳、夹持不稳；这一自述与所见滑脱相符，但不足以区分接触布局、肌肉力分配或跟踪反馈的具体责任，根因停在控制。"
+   }
+  ],
+  "gap_zh": "无状态日志；两视角中侧视部分被起始柱遮挡，右臂100步累计接触的首次达标时刻由俯视接触和持物过程估计，不能精确恢复每个控制步的接触标签。评分器未报告DAG所列五项原生指标，均填null；没有本次交付轨迹的指尖力与目标/实际关节跟踪序列，故无法确定控制根因的子类。"
+ },
+ "summary_zh": "肌肉右手夹住并抬起物体，按DAG的累计接触历史规则估计在视频约 12.5 s 达标，最终保留1/3节点；假肢手交接和目标柱放置均未达到。右手在视频约 16.3 s 未保持住物体，物体滑脱下落，交付回放在471个控制步结束并判失败。缺少逐步接触、力和关节跟踪日志，达标时刻为估计，根因仅归入控制。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注与视频一致：右手自视频约 8 s 夹住长方体、13–16 s 抬起、约 16.3 s 滑脱；节点1估计达标时刻（约 12.5 s）在约1 s误差内合理，交接与放置未达成，断点、现象（没保持住）和根因（控制，停在大类）均有证据支持。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:40:27Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--ahkgq8f/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--ahkgq8f/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "96fa2810a9e2d24b6a4673c7cc1f6fca4d7382541625cc4776c2743d75a67a63",
+   "video_sha256": "881dbc4fcab865b390e5376a7bc1f95ac6cb545288763daa048c4983d834c6e7",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--t8bv7go.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--t8bv7go.json
new file mode 100644
index 0000000000000000000000000000000000000000..440012d703dfd9acddcebceb5773df0251503f45
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--t8bv7go.json
@@ -0,0 +1,284 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-privileged--t8bv7go",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 1000,
+   "sim_steps": 1000,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 29,
+   "agent_resets": 79,
+   "agent_sim_steps": 19832,
+   "agent_records": 9,
+   "agent_replays": 1,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.577,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.25,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-privileged--t8bv7go/replay.mp4",
+  "duration_s": 40.04,
+  "fps": 25.0,
+  "frames": 1001,
+  "sha256": "0df7f7acf7e117d57bcfff8294e3d7a39828886e695836d1ffba13e84d7edc1f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [
+     [
+      6.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 6.5,
+      "log_step": 52,
+      "text_zh": "观察：视频1.25–7.5 s，肌肉右手及腕部贴近并接触红边长方体，物体在起始柱面上倾倒、滑动；视频8.5–16 s，右手移走而物体留在原柱。log step 52 是生成并保存最终交付动作的运行，脚本直接统计其接触历史，输出 myo 122、pros 0、goal 0；log step 53 对该文件回放得到 deterministic=true、1000步失败，与视频相符。"
+     },
+     {
+      "t": 6.5,
+      "log_step": 52,
+      "text_zh": "推断：按本DAG的实际验收，肌肉臂任一部位累计接触122步已超过100步，因此该历史事件达成并保持为真；不把它解释为成功夹起物体。视频共1001帧、25 fps，对应交付1000控制步及初帧，每控制步约0.04视频秒；log step 52 的控制步200物体仍在柱上，与视频8.0 s核对相符。达标时刻估在视频6.5 s附近，误差约1 s；日志只有总计，不能精确重建第100个接触步。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频1.25–8.0 s，肌肉右手/腕部与长方体侧面接触，把物体沿起始柱顶推移、压倒；未见手指从相对两侧稳定夹住物体，故不标抓握。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 19,
+      "log_step": 52,
+      "text_zh": "观察：视频8.5–16 s，红边长方体留在起始柱，肌肉右手空手移向两臂之间；视频18–26 s，白色假肢手伸到空交接位并合拢，物体没有进入白手。log step 52 对实际保存运行的接触历史统计为 pros 0，问题为PROSTH_SHORT。"
+     },
+     {
+      "t": 40,
+      "log_step": null,
+      "text_zh": "观察：视频27–40.0 s，长方体始终留在起始柱，白手移向另一侧后没有带回物体。成功参考视频18–20 s可清楚看到长方体从肌肉手转到白手；本次同类视角没有该物体转移，故交接未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频19–26 s，白色假肢手在空交接位张合，未接触仍在起始柱的长方体。候选“空抓尝试”：没有实际包围或夹住物体，不能写抓握，也没有两手共同接触物体；现有方式无对应值。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 30,
+      "log_step": 52,
+      "text_zh": "观察：视频27–36 s，白色假肢手空手从交接区移向目标柱一侧；长方体仍可在起始柱上辨认，没有随白手移动或落上另一个目标柱。log step 52 的交付运行接触历史统计 goal 0。"
+     },
+     {
+      "t": 40,
+      "log_step": null,
+      "text_zh": "观察：视频39.0、39.25、39.5、39.75、40.0 s，长方体持续留在肌肉臂旁的起始柱面，未出现在目标柱顶；目标放置关系从未建立。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频27–36 s，白色假肢手执行向目标柱一侧移动的空手放置动作，长方体留在起始柱。候选“空手放置尝试”：手中没有物体，无实际抓握或释放接触，现有接触方式没有对应值。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_handed_to_prosthesis",
+   "t": 8.5,
+   "log_step": 52,
+   "where": "操作中",
+   "evidence_zh": "观察：视频7.5–9.5 s，肌肉右手开始离开而长方体落平并留在起始柱；视频12–26 s，两臂继续执行交接位置的动作，白手在空处合拢，未承接物体，此后也未修复。log step 52 生成并保存的1000步轨迹中，肌肉臂历史接触122步达标，假肢接触0步，最终PROSTH_SHORT；log step 53确定性回放同一文件仍失败。推断：按评分器口径第一历史事件已达成，首个始终未完成的操作步骤是交给左手；视频8.5 s作为物体被留在起始柱的导航锚点。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 8.5,
+    "log_step": 52,
+    "evidence_zh": "观察：视频1.25–8.0 s，肌肉右手/腕部贴靠红边长方体并使其在起始柱顶倾倒、滑动；视频8.5–14.0 s，右手抬起并移向交接区，物体仍留在起始柱，未见随右手离柱搬运。log step 52 是生成并保存交付轨迹的运行，控制步300–900的obj位置约为[-0.299,-0.294,1.103]，而palm位置继续变化，累计myo接触122步、pros接触0步。推断：右手虽有推靠接触，却未建立将物体带到交接位所需的夹持或承载接触，导致交接失败；视频19–26 s白手空合拢仅作后续旁证，不作为独立的接收抓取失误。现有证据未显示物体曾被右手持起后再滑落，故保留“没建立接触”，不改为“没保持住”。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "视频 / 日志",
+    "evidence_zh": "推断：所交付承托方案没有形成可随手搬运的承载接触。视频1.25–8.5 s，肌肉手/腕贴靠长方体并将其沿柱面推移，随后手上移、移出而物体留柱；视频12–26 s因此执行空手交接。log step 34 的控制器明确采用自由下垂后把手当托盘的方案，log step 43 将其改为固定初始手姿缓慢上提，log step 51将该策略内联到最终生成脚本，log step 52显示物体在控制步300后停在约[-0.299,-0.294,1.103]，而手继续移向交接区。该接触布局未托住或夹住物体；没有证据把失败进一步归到某个力控增益或逆解误差。成功参考展示任务可完成，但未验证本次冻结初态下的具体替代接触方案。"
+   }
+  ],
+  "gap_zh": "未提供状态日志或逐控制步接触历史；肌肉臂100步达标时间只能由早段画面和最终累计122步估计，视频6.5 s锚点可能相差约1 s。接触标签覆盖整臂，画面不足以逐项区分腕部与各指接触。原生评分reward字段未报告DAG列出的五项指标，native_metrics均留null；log step 52的myo/pros/goal是最终生成脚本的直接诊断，用于节点证据，不冒充reward指标。未验证本冻结初态的替代夹持方案。",
+  "unmet": {
+   "object_held_by_myo": true,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  }
+ },
+ "summary_zh": "本次交付完成肌肉臂累计接触要求（log step 52：122步），按实际评分口径计1/3节点；视频中物体只在起始柱上被推移，没有被带到交接位。断在交给左手：承托/夹持布局未能携物，白手在空处合拢，假肢接触及目标柱接触均为0；视频40.0 s物体仍留起始柱。缺少逐步接触历史，首节点达标时刻只能估计，不能把接触计数达标等同于稳定持起。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 5,
+    "summary_zh": "节点值与断点基本与视频及log step 52/53一致（物体始终留在起始柱，白手空合拢，假肢接触0）。唯一问题：现象“没建立接触”落在白手上，而根因与证据说明失败源于右手没有把物体带到交接位，现象与根因对象不一致。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点值、断点和现象与视频及log step 52/53一致：物体全程留在起始柱，肌肉手只推移，假肢接触0；第一节点按评分器接触计数(122)判达成，已在证据中标为推断。未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:10:23Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--t8bv7go/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--t8bv7go/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--t8bv7go/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--t8bv7go/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "7de4f03b070a6e2e56daba31148187658debb8f1ecf0f244bc4a6edf49529d91",
+   "video_sha256": "0df7f7acf7e117d57bcfff8294e3d7a39828886e695836d1ffba13e84d7edc1f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--vfsmu3q.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--vfsmu3q.json
new file mode 100644
index 0000000000000000000000000000000000000000..a4c4685e162963a0ad213cc6ea1a57fd35e4f69a
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--vfsmu3q.json
@@ -0,0 +1,266 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-privileged--vfsmu3q",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 960,
+   "sim_steps": 960,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 149,
+   "agent_resets": 202,
+   "agent_sim_steps": 19515,
+   "agent_records": 7,
+   "agent_replays": 3,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 11.513,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.246,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 2,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_oom": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-privileged--vfsmu3q/replay.mp4",
+  "duration_s": 38.44,
+  "fps": 25.0,
+  "frames": 961,
+  "sha256": "09d683fc1b9ca10d1052541779952bdc0d863c174f69f0ffe8a90bb587c141d4"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 1.4,
+      "log_step": null,
+      "text_zh": "观察：视频约1.4 s起，肌肉右手掌指贴住红边长方体上缘，视频2.8–9.6 s手指弯曲并持续压推物体，物体沿起始柱顶移动，未被稳定抬离柱面。"
+     },
+     {
+      "t": 5.5,
+      "log_step": 8,
+      "text_zh": "推断：按DAG及log step 8的评分代码，此节点接受MYO累计100个控制步接触，不另要求抬升。最终交付有960步、录像有961帧且为25 fps，每步约0.04视频秒；log step 102、103、108给出的最终阶段长度对应假肢手约视频34.8 s开始张手，已与视频35.0–36.0 s画面核对。结合视频1.4–9.6 s持续贴靠及随手推移，估计约视频5.5 s累计达标，首次达标时刻约有1 s不确定性，历史事件此后保留。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "肌肉右手掌指从上缘和侧面压住红边长方体，并沿起始柱顶推移；未见对向夹紧后承重抬起。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 14.4,
+      "log_step": null,
+      "text_zh": "观察：视频9.6–16.3 s，白色假肢左手靠近起始柱、弯曲手指，但手与红边长方体未会合；右侧画面中物体一直留在起始柱，左侧画面可见白手在物体另一侧较远处闭合。"
+     },
+     {
+      "t": 22.56,
+      "log_step": 108,
+      "text_zh": "观察：视频18.2–30.0 s白色假肢手离开时手中没有红边物体，起始柱原位仍有该物体，视频38.4 s亦如此。最终交付的log step 108显示960步实时与回放终态哈希相同且solved=False；该日志只用来确认交付身份，不把其他会话试验当作本次交接。"
+     },
+     {
+      "t": 16.3,
+      "log_step": null,
+      "text_zh": "推断：白手未建立承接接触，也未使物体随其移动，因而此节点未达到；不是已经接到后滑落。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "压/擦",
+      "note_zh": "交接尝试时肌肉右手仍贴压红边长方体，假肢左手同时靠近并空合手指；仅右手实际接触物体，未形成双手抓握或左手承接。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 30.0,
+      "log_step": 108,
+      "text_zh": "观察：视频22.6–30.0 s假肢左手空手向另一侧移动，红边长方体仍留在起始柱；最终交付log step 108的搬运后诊断也显示物体位置约为[-0.3359,-0.2051,1.134]，假肢掌部已经向另一侧移动。"
+     },
+     {
+      "t": 38.4,
+      "log_step": null,
+      "text_zh": "观察：视频31.0–38.4 s，肌肉右手逐渐张开，红边长方体留在同一起始柱顶；未见物体被送到与起始柱分开的目标柱。末秒按约0.12 s间隔检查至视频38.4 s，未出现交接或目标柱放置。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "候选：空手搬放。白色假肢左手在未接到红边长方体时执行朝目标方向移动及张手撤离的动作，没有对物体的搬放接触；现有抓握、托/舀等值均不适合这次实际空手动作。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_handed_to_prosthesis",
+   "t": 14.4,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约14.4 s，假肢左手在起始柱另一侧合拢，红边物体仍在肌肉右手附近的起始柱顶，没有进入白手；视频18.2–30.0 s白手空手离开，直至视频38.4 s交接仍未修复。右手节点按累计接触规则已达到，最先未修复的是左手接收。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 14.4,
+    "log_step": null,
+    "evidence_zh": "观察：视频13.4–16.3 s，白色假肢手弯指位置与红边物体分离；视频18.2–22.6 s假肢手移动后物体仍在起始柱。所需的左手承接接触从未建立，未见一次持有后丢失。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "视频 / 日志",
+    "evidence_zh": "推断：视频2.8–14.4 s肌肉右手以掌指压推物体，未形成承重夹持或把物体送到白色假肢手闭合处；视频14.4–16.3 s两手的作用位置分开，承载与交接接触布局未建立。log step 109中agent自述多种握法均把物体推开、无法抬起，是会话层面的辅助证据；未把其可达性判断或早期控制错误当成本次已证实原因。"
+   }
+  ],
+  "gap_zh": "无状态日志，也没有最终交付的逐步接触集合或累计接触计数；右手100步达标及约视频5.5 s的时刻由画面接触持续性估计，不能精确恢复。log step 109先称右手持续接触超过100帧、后又称两臂条件均未满足，不能据此替代画面判定。白手闭合处部分被肌肉臂遮挡，但起始柱物体始终未被带走、后续白手为空的画面足以判交接失败；无法进一步量化接触力或区分具体逆解与跟踪误差。评分器未报告DAG五项原生指标的数值，均保留null。",
+  "unmet": {
+   "object_held_by_myo": true,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  }
+ },
+ "summary_zh": "本次交付中肌肉右手持续压推起始柱上的物体，按DAG的累计接触验收估计达到右手节点（约视频5.5 s），但未把物体稳定抬起。白色假肢手靠近、空合后空手移开，交接失败且未修复，物体到视频38.4 s仍在起始柱，最终1/3；可见的接触与承载布局不足支持控制层面的抓取和接触设计问题。无状态日志及逐步接触计数，右手首次达标时间存在约1 s不确定性，原生指标未报告。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "右手节点按累计接触规则估计达成（约视频5.5 s）、交接未达成、目标柱未达成，断点在交接（操作中）、现象没建立接触、根因抓取和接触设计，均与视频和log一致；未发现改变读者结论的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:46:46Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--vfsmu3q/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--vfsmu3q/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "e513e7d0aba79491cc50fc620d525d1e78e8c7c87ab6010b3921badc5397469d",
+   "video_sha256": "09d683fc1b9ca10d1052541779952bdc0d863c174f69f0ffe8a90bb587c141d4",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--xxpz8gf.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--xxpz8gf.json
new file mode 100644
index 0000000000000000000000000000000000000000..67c0bfdc5e4776fe849efddbbd979443b167828d
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--xxpz8gf.json
@@ -0,0 +1,255 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-privileged--xxpz8gf",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 1000,
+   "sim_steps": 1000,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 441,
+   "agent_resets": 381,
+   "agent_sim_steps": 99321,
+   "agent_records": 0,
+   "agent_replays": 7,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 11.888,
+   "peak_processes": 11,
+   "peak_disk_gb": 1.013,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 2,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_oom": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-privileged--xxpz8gf/replay.mp4",
+  "duration_s": 40.04,
+  "fps": 25.0,
+  "frames": 1001,
+  "sha256": "88c70b129aa4784bf3a49dd1a524a738bc9e1e236aaf67d4f12cd8b982f838c6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 11.8,
+      "log_step": null,
+      "text_zh": "观察：视频 0–40.0 s 的两视角中，肌肉右手反复偏离起始柱；视频 11.6–12.2 s 指尖从盒侧经过，未形成对向包围，随后空手离开。视频 24.4–25.6 s、32.4–35.8 s 再次经过时，盒仍由起始柱承托，没有随手搬走。"
+     },
+     {
+      "t": 35.8,
+      "log_step": 273,
+      "text_zh": "推断：逐段检查近物画面后，右手仅有短暂擦碰机会，未达到累计100控制步的接触要求，故本历史事件不记达成；并非由零分或未抬高单独判定。交付文件在 log step 273 的两次回放均为1000步、确定性失败；1001个视频帧对应初态及1000动作，25 fps下100控制步约占视频4 s，已用视频4.4 s附近的合指动作核对 log step 103 的110步合指日程。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "推断：视频11.6–12.2 s肌肉右手指尖沿红边盒侧短暂擦推；没有看到拇指与其他手指对向夹住盒，不能标为抓握。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 18,
+      "log_step": null,
+      "text_zh": "观察：视频 14–21 s 白色假肢手在另一侧空手运动，右手也为空手；视频 24–26 s 和32–35.8 s 盒再次进入画面时仍在原起始柱上，未见白手包围或承接盒并带其移动。"
+     },
+     {
+      "t": 40,
+      "log_step": null,
+      "text_zh": "推断：结合视频 0–40.0 s 两视角的前后位置，未发生物体向假肢臂的交接，也没有足量假肢接触的可见过程；末秒白手仍空手，故该事件未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频14–21 s可见白色假肢手执行空手动作，但没有可标注的盒接触；不把空手合指或右手在别处擦盒记为双臂抓握。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 35.8,
+      "log_step": null,
+      "text_zh": "观察：视频 32.4–35.8 s 盒仍在起始柱上，分开的目标柱未承接盒；视频 39.0–40.0 s 右手和白色假肢手均未持盒。全片没有盒从起始柱被送往目标柱的过程，故放置条件始终不成立。"
+     },
+     {
+      "t": 40,
+      "log_step": 273,
+      "text_zh": "观察：log step 273 对最终交付文件两次回放的success均为false，instance_ok和deterministic均为true；这与视频未完成放置一致，输出未报告goal_dist、goal_touch或末尾目标柱独占接触计数。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频后段白色假肢手为空手姿态，没有看到它持盒放置；无法为脚本的空手放置动作赋予物体抓握接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_held_by_myo",
+   "t": 1.5,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0–1.5 s 肌肉右手从盒上方移向远离起始柱的位置，手中没有盒；视频 9.5–13 s、24–26 s 和32–35.8 s虽再次经过起始柱，仍未完成足量接触及持物搬运。这是最先未修复的步骤，后面的白手空手动作没有修复它。"
+  },
+  "unmet": {
+   "object_held_by_myo": false,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.5,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.4–1.5 s 肌肉右手迅速远离仍在起始柱上的盒；视频 2–8 s、14–23 s 多数时间手在空处运动。视频 11.6–12.2 s 即使扫到盒侧也未形成包围，之后空手离开；未看到盒随右手到达交接位置。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 156给出的控制器用掌心雅可比生成关节目标，再经肌肉激活映射执行；log step 169、171的独立跟踪测试中，掌心目标为[-0.522,-0.226,1.292]，实际终点却约为[0.03,0.20,1.73]，误差从约0.115 m增长至约0.83 m。log step 211随后用同一gen.py阶段脚本生成最终1000步文件，log step 273验证该文件确定性失败；评分视频0–8 s及14–23 s的空手大幅偏离与此控制能力缺陷一致。测试属于会话根因证据，不用作节点进度或接触计数；不足以进一步区分反馈控制与接触设计。"
+   }
+  ],
+  "gap_zh": "未提供状态日志，评分器也未输出逐步接触历史和各原生指标，右手累计100控制步是否达标依据近物片段作视觉推断，不能恢复精确计数或接触身体标签；擦推接触也是结合前后画面的推断。末帧盒不在视野内，但视频35.8 s仍在起始柱、随后双手均空手，没有到达目标柱的过程。agent在log step 280自述“扫下起始柱并掉落”，与评分视频中物体反复仍在柱上的观察不一致；不采纳该掉落叙述，log step 273的统一失败原因也不能单独证明掉落。"
+ },
+ "summary_zh": "评分轨迹未完成右手足量接触、交给左手或放上目标柱，最终进度为0/3；肌肉右手反复偏离盒，仅短暂擦过，白色假肢手始终未承接盒。断点在右手持起步骤，现象为位姿不准，会话测试与评分视频共同支持控制问题；无状态接触历史，右手计数及擦推接触为视觉推断，agent关于掉落的最终自述不作为评分轨迹事实。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "三个节点均未达成、断点为右手持起、现象位姿不准、根因控制，与视频0–40 s两视角所见（手多数时间在空处、盒在起始柱附近未被携走）一致；未发现需要改动的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:42:22Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--xxpz8gf/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--xxpz8gf/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "cdf416aa6147381788ac639e8c8b87be64444253baf21749fbc2c1b29eae4a22",
+   "video_sha256": "88c70b129aa4784bf3a49dd1a524a738bc9e1e236aaf67d4f12cd8b982f838c6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--zrrewr9.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--zrrewr9.json
new file mode 100644
index 0000000000000000000000000000000000000000..1fe87644dfc34c9bb39a58b687de333c19c24cb0
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-privileged--zrrewr9.json
@@ -0,0 +1,277 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-privileged--zrrewr9",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 350,
+   "sim_steps": 350,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 254,
+   "agent_resets": 6,
+   "agent_sim_steps": 22156,
+   "agent_records": 3,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.806,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.275,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-privileged--zrrewr9/replay.mp4",
+  "duration_s": 14.04,
+  "fps": 25.0,
+  "frames": 351,
+  "sha256": "ccfdfc762fca1e96e30452e8e4f230ed38fbbd886db77c0e627d7227a4210ab2"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.7,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 2.4,
+      "log_step": null,
+      "text_zh": "观察：已检查视频 1.2–4.4 s 的密集画面；视频约 1.7 s 起，肌肉右手指端接触红边灰色盒并使盒体倾斜，视频 2.4–4.4 s 盒体继续位于手指下方，随后保持接触并随手移动。该节点按实际评分的累计接触判定，不另要求先明显抬高。"
+     },
+     {
+      "t": 5.7,
+      "log_step": 411,
+      "text_zh": "推断：log step 411 对交付文件逐动作重放，统计肌肉臂接触共238个控制步，超过100步；log step 407、418 对同一350步文件回放的终态哈希一致且 deterministic=true。视频约 1.7 s 起已见右手接触并压斜盒体，之后画面支持持续接触，因此以此作为累计起点估计：100个控制步约为4.0视频秒，首次达标约在视频 5.7 s，估计不确定性约±0.7视频秒。350个动作对应351帧、25 fps，即每控制步约0.04视频秒；log step 405 的交付脚本从控制步130开始向滑移目标插值，对应视频约 5.2 s，已与视频 5.0–5.6 s 盒体转动并向柱缘移动核对。未换算agent工作时间；总计数不证明每步均连续接触。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "肌肉右手先压倒盒体，随后拇指与其余手指夹住红边灰色盒，沿起始柱面向外移动；没有把柱面上的移动误作额外抬升验收。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 12.8,
+      "log_step": 411,
+      "text_zh": "观察：视频 12.0–14.0 s，两手在起始柱外侧会合，盒体下缘接触白色假肢手伸出的指尖；白手没有合拢包住盒体，也没有携盒离开交接处。log step 411 对交付文件逐动作重放，假肢臂累计接触只有33个控制步，未达到100步；这是已由log step 407、418确认确定性的交付文件，非较早试验的计数。"
+     },
+     {
+      "t": 14.0,
+      "log_step": 410,
+      "text_zh": "观察：视频 13.28–14.0 s 盒仍斜悬在起始柱旁和白手指尖处。log step 410 重放同一交付文件，终态为350步、obj_pos=[-0.185,-0.266,1.05]、touching_body=[0,1,0,0,0]、success=false、done=false，哈希与交付回放一致；末帧存在假肢接触不等于累计100步交接达标。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "抓握",
+      "note_zh": "视频末段肌肉右手夹持盒体送向白色假肢手，两手在盒体处会合；此处抓握指肌肉右手，白手未形成包围夹持。"
+     },
+     {
+      "who": "双臂协作",
+      "how": "托/舀",
+      "note_zh": "白色假肢手保持手指伸直，在右手送出的盒体下缘以指尖承接；可见接触限于指尖附近，未托入掌内。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 14.0,
+      "log_step": 410,
+      "text_zh": "观察：已检查视频全程以及视频 13.0–14.0 s 的密集末段，盒体从起始柱向白手移动后始终停留在交接侧，未被白手带到另一目标柱顶，也未出现放柱后双手离开的状态。log step 410 的交付回放终态仍处于交接坐标，success=false；成功参考视频约24–31 s中另柱附近的放置过程仅用于辨认目标位置，不作为本次完成的证据。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_handed_to_prosthesis",
+   "t": 13.3,
+   "log_step": 410,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 12.8–14.0 s，肌肉右手把盒送出起始柱，白手伸指在盒底承接，但盒体只落在伸直的指端附近，未进入掌内，未随白手搬走。log step 411 对交付350步文件的回放给出假肢臂33步接触，未达该步100步验收；log step 418 两次确定性回放均success=false，且没有后续修复。"
+  },
+  "unmet": {
+   "object_held_by_myo": true,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 13.3,
+    "log_step": 410,
+    "evidence_zh": "观察：视频 12.8–14.0 s，白色假肢手掌在盒体侧下方，伸直的指尖只碰到盒体下缘，盒体在柱缘外向下倾斜，始终没有送入白手掌内的承接区域。log step 410 的交付文件终态仍仅有假肢臂接触；这里是交接承接位置偏差，末帧没有显示物体已经掉到地面。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 405 的交付生成脚本在接收阶段持续使用固定q_prosth_bridge，全部350个控制步均令假肢手指命令p_fingers=-1，未在盒到达后加入合指或继续托入掌内的动作；视频 12.8–14.0 s 也显示白手仅用伸直指端承接、盒体位于掌外。该接触布局未把物体送入足够的承载区域，交付回放的假肢接触仅33步（log step 411），因此归因于接触设计；没有在本次交付中验证替代接收布局。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志或交付轨迹的逐步接触计数；依据视频约 1.7 s 起的可见接触及后续夹持、每控制步约0.04视频秒，将右臂累计100步的首次达标估计为视频约 5.7 s，估计不确定性约±0.7视频秒。日志可确认最终238/33步，但不能精确恢复每次接触的起止或排除短暂中断。评分器reward未报告DAG五项原生指标，native_metrics均保留null。视频在白手指尖承接中结束，不能把会话其他测试的后续掉落当成本次评分事件。"
+ },
+ "summary_zh": "本次交付中，肌肉右手夹持并将盒体移出起始柱，按累计接触验收完成右手节点，最终进度1/3；白手仅在指尖处承接，交付回放接触33步，交接和目标柱放置均未达标。断点在交给左手：盒体未送入掌内，固定接收姿态与持续张开的手指缺少进一步承接布局。没有逐帧接触日志，按视频约 1.7 s 起接触并累计100控制步推断，右手达标约在视频 5.7 s（估计不确定性约±0.7视频秒），其他会话试验中的掉落未计入本次评分轨迹。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点2/3、断点（交给左手·操作中）、位姿不准与接触设计的标注与视频和 log 一致；主要问题是右手节点达标时刻 7.5 s 的推算起点错误，视频显示右手自约 1.7 s 起已持续接触盒体，100 步累计应在约 5.5–6 s。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "第一轮已修正右手节点时刻；节点 1/3、断点（交给左手·操作中）、位姿不准与接触设计均与视频和 log 一致，未发现新错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:11:43Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--zrrewr9/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--zrrewr9/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--zrrewr9/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-privileged--zrrewr9/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "12c572cb4c56beac43446f9705b9ea260684f222232f45185ce6c4487509c3f1",
+   "video_sha256": "ccfdfc762fca1e96e30452e8e4f230ed38fbbd886db77c0e627d7227a4210ab2",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--9hfsnqg.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--9hfsnqg.json
new file mode 100644
index 0000000000000000000000000000000000000000..b9c14699792e30513a36a8d5858e73e9fa79e4a0
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--9hfsnqg.json
@@ -0,0 +1,273 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-standard--9hfsnqg",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 919,
+   "sim_steps": 919,
+   "physics_steps": 4595,
+   "steps": 1909,
+   "calls": 35,
+   "episodes": 4,
+   "resets": 3,
+   "reset_budget": 10,
+   "sim_steps_total": 1909,
+   "elapsed_s": 2700.9,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 4.075,
+   "peak_processes": 34,
+   "peak_disk_gb": 0.402,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-standard--9hfsnqg/replay.mp4",
+  "duration_s": 36.8,
+  "fps": 25.0,
+  "frames": 920,
+  "sha256": "637bc841d93aa9f147b055e3ab5c8f1fb01c098d5b9ceb91ed0beab4517d2add"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 3.6,
+      "log_step": null,
+      "text_zh": "观察：视频 3.0、3.6、5.0、7.0 s 的俯视画面显示肌肉右手位于红边长方体一侧，拇指贴近其上缘，其余手指在柱顶外侧；另一视图的接触区域被起始柱遮挡，不能分辨拇指与物体是否持续实接触。"
+     },
+     {
+      "t": 14.0,
+      "log_step": null,
+      "text_zh": "观察：视频 8.0–14.0 s 右手合指并移开，视频 14.0、18.0、22.0、24.0 s 红边长方体仍在起始柱上，右手中没有物体。物理持起没有发生；但本节点按肌肉臂累计接触至少100个控制步验收，并不要求抬升，故不能仅凭空抓把接触计数判为不达标。"
+     },
+     {
+      "t": null,
+      "log_step": 79,
+      "text_zh": "观察：log step 79 将动作序列发送到重置后的服务，服务状态为 episode=4、control_steps=919；log step 81 仍为同一局同一控制步数。log step 78 的 myo=586、pros=184 来自本地仿真，不是该受评局的接触计数，不能据此判本节点达标。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "肌肉右手在视频 3.0–12.0 s 于物体旁合指；拇指是否实接触、如何施力不能由柱体遮挡后的两视图可靠确认，未见包住物体并持起，不标为实际抓握。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 24.0,
+      "log_step": null,
+      "text_zh": "观察：视频 14.0、18.0、22.0、24.0 s 物体留在起始柱，肌肉右手已空手移向两臂之间；视频 26.0–32.0 s 白色假肢手掌朝上来到右手下方，两手会合时均未携带红边长方体。"
+     },
+     {
+      "t": 36.76,
+      "log_step": null,
+      "text_zh": "推断：视频 33.0–36.76 s 白手空手离开，未见物体进入其手掌或随其移动；结合视频 24.0 s 仍在起始柱的物体及此前右手空手搬运，假肢臂没有获得该物体的100步接触，节点未达成。成功参考录像在参考视频 18.0–21.0 s 可见红边长方体在两手间转移，本局没有该可见变化。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频 26.0–32.0 s 肌肉右手与白色假肢手执行空手会合；没有物体进入两手之间，未建立可标为抓握或托/舀的物体接触。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 24.0,
+      "log_step": null,
+      "text_zh": "观察：视频 24.0 s 红边长方体仍在起始柱；视频 26.0–33.0 s 白色假肢手执行承接、移动及撤手动作时，掌中没有物体。"
+     },
+     {
+      "t": 36.76,
+      "log_step": 79,
+      "text_zh": "推断：已逐段检查视频 0–36.76 s，含视频 35.8、36.0、36.2、36.4、36.6、36.76 s 的末段画面，没有物体被持离起始柱或送至目标柱的过程；所以未形成目标柱放置关系。评分器 success=0 与此一致，但不是仅凭总分判该节点为假。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频 32.0–36.76 s 白色假肢手空手移开，没有可分类的手与物体的搬运或释放接触；后续空手动作不能记为实际抓握放置。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_held_by_myo",
+   "t": 12.5,
+   "log_step": 79,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 8.0–12.5 s 肌肉右手合指后开始离开起始柱，视频 13.44–14.0 s 手与红边长方体明显分离，物体仍在柱面；视频 18.0–24.0 s 也没有补抓，之后两手执行空手交接。最早未修复的是右手持起的实际操作；该节点的历史接触阈值本身仍因计数缺失保留未知。"
+  },
+  "unmet": {
+   "object_held_by_myo": null,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 12.5,
+    "log_step": null,
+    "evidence_zh": "观察：视频 8.0–14.0 s 右手在物体旁屈指后空手移开，红边长方体没有随手离开起始柱；视频 18.0–24.0 s 物体仍在柱面。没有建立能带起物体的对向夹持；这不排除拇指曾有轻微擦碰，也不是已持起后滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 53 的服务控制脚本将本地计算的80执行器命令按100行一批发送，循环中没有使用摄像头或本体反馈修改抓取，只在全部命令后取图；log step 79 实际运行该脚本形成 episode 4。视频 12.5–24.0 s 的空手右臂仍继续搬运，视频 26.0–32.0 s 白手继续空接，说明受评控制序列没有在抓取未建立时纠偏。log step 82 中 agent 也自述本地成功抓取没有在服务复现；具体是接触几何还是执行跟踪差异，现有证据不能再细分。"
+   }
+  ],
+  "gap_zh": "输入没有状态日志，评分器没有报告两臂累计接触计数及各项原生指标。视频 3.0–12.0 s 的右手指与物体上缘在俯视图中紧邻，另一视图被柱体遮挡；无法可靠区分持续接触与间隙，故右手100步历史接触事件为未知，而不是用未抬升代替评分规则判假。左手交接与放置可由物体始终留在原处、两手空手执行后续动作判未达成。log step 78 的接触计数和掉落是本地测试，不用于本局进度或现象；log step 82 猜测物体质量、摩擦差异未获独立验证，未作为环境根因。"
+ },
+ "summary_zh": "受评的是 log step 79 运行的最后一局 episode 4：右手在起始柱旁合指后空手移走，物体没有被持起，随后假肢手执行空手承接与撤离，交接和目标柱放置均未达成。断点为右手持起操作，现象是未建立有效夹持；脚本按预计算命令继续执行而没有根据抓取失败纠偏，根因标为控制。没有受评局的接触计数，且关键近接触面被柱体遮挡，故按实际100步接触验收的右手事件保留未知，确认完成节点为0/3。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "逐帧核对后，右手在视频 2–12 s 于物体旁合指、14 s 起空手离开、物体留在起始柱，与文件一致；两个下游节点为假，评分器 success=0 且无状态日志，对第一个节点标 unknown 有依据；断点、现象（没建立接触）与根因（停在“控制”）有证据支持，未发现需要改动的问题。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:41:32Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--9hfsnqg/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--9hfsnqg/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6d41a14e3c278ff5326ff58a2507d187a92d1598cbd69da52878fe515e45fcd2",
+   "video_sha256": "637bc841d93aa9f147b055e3ab5c8f1fb01c098d5b9ceb91ed0beab4517d2add",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--d2hna7y.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--d2hna7y.json
new file mode 100644
index 0000000000000000000000000000000000000000..97a366add67cfda1580c6e78c3c5129d6e1e6884
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--d2hna7y.json
@@ -0,0 +1,282 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-standard--d2hna7y",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 450,
+   "sim_steps": 450,
+   "physics_steps": 2250,
+   "steps": 460,
+   "calls": 19,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 460,
+   "elapsed_s": 2601.0,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.754,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.389,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-standard--d2hna7y/replay.mp4",
+  "duration_s": 18.04,
+  "fps": 25.0,
+  "frames": 451,
+  "sha256": "8922835744863e63ef6805b2c29bef1c8197f7423a5be4ef51c1b8bb6bc9860a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 1.2,
+      "log_step": null,
+      "text_zh": "观察：视频0.8–2.2 s，肌肉右手弯曲手指接触红边盒体，盒体随手指运动沿起始柱面转动、滑移；视频4.8–18.0 s，盒体停在起始柱缘，仍贴靠肌肉臂腕部，未完成脱柱搬运。"
+     },
+     {
+      "t": 5.5,
+      "log_step": 344,
+      "text_zh": "推断：按DAG的实际验收，肌肉臂任一部位累计接触100个控制步即可，不另要求抬离柱面。视频约0.8 s开始接触，随后手指至腕部的接触延续，约视频5.5 s已累计足量接触（达标时间为视觉估计，误差约0.5 s）；log step 344的评分局提交450控制步，与451帧、25 fps、视频18.04 s对应，100步约占视频4 s，已核对视频18.0 s的末态。未用agent工作时间换算。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约0.8–2.2 s，右侧肌肉手弯指包夹红边盒体；盒体仍由起始柱支承，未见稳定抬离。"
+     },
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频约4.8–18.0 s，盒体由起始柱缘支承并贴靠右侧肌肉臂腕部；此处为腕部抵压，不能据此称为持续抓握。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 6.4,
+      "log_step": 344,
+      "text_zh": "观察：视频3.6–6.4 s，白色假肢手伸向右手下方；视频6.4–18.0 s，两手停在起始柱边，盒体仍贴靠肌肉臂腕部和起始柱缘，没有随白手离开。白手指尖与盒体之间的局部区域被肌肉手腕及柱体遮住，不能逐帧核实微小接触。"
+     },
+     {
+      "t": 18.0,
+      "log_step": 267,
+      "text_zh": "观察：log step 267的415步肌肉接触、128步假肢接触来自agent自行建立的本地MuJoCo测试；log step 273的假肢累计达标也来自本地测试，并非评分局服务读取。log step 345与346分别记录远端及本地末态，肘屈曲读数分别为1.2775与1.3959 rad，不能视为同一轨迹。"
+     },
+     {
+      "t": 18.0,
+      "log_step": 358,
+      "text_zh": "推断：视频明确未展示物体由白手接走，但DAG实际只验收假肢臂累计接触100控制步，缺少搬运不等于该计数未达标；局部接触被遮挡且没有评分局接触历史，因此保留未知，不采用agent在log step 358引用的本地128步作为节点值。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频6.4–18.0 s的交接尝试中，盒体仍抵在肌肉右臂腕部和起始柱缘，右侧作用端没有携物离开。"
+     },
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频5.8–7.8 s及末段，白色假肢手伸展指尖靠近物体所在的右腕下方，接触界面被右腕及柱体遮挡，无法确认白手是否有轻触及其方式；未见白手包夹并携走盒体。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 18.0,
+      "log_step": 357,
+      "text_zh": "观察：视频0.0–18.0 s中，盒体从起始柱面移到同一柱的边缘后停住；视频17.0、17.2、17.4、17.6、17.8和18.0 s均未出现白手将盒体送上另一根目标柱并离开的动作。log step 357的评分局服务状态仍为control_steps=450、done=False；评分器最终success=0。"
+     },
+     {
+      "t": 6.4,
+      "log_step": 354,
+      "text_zh": "观察：log step 344提交的动作及log step 354保存的脚本只让假肢手插值到交接位并维持，没有目标柱运输或释放阶段；视频中也未发生向另一根目标柱的携物运动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_handed_to_prosthesis",
+   "t": 6.4,
+   "log_step": 344,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约6.4 s，白色假肢手到达肌肉右手下方后未包住或接走盒体；盒体继续留在起始柱缘和右腕处，直至视频18.0 s。log step 344是评分局唯一450步动作提交，随后log step 355、357均显示仍停在control_steps=450。推断：最早未修复的物理操作停在交接承接处；假肢累计接触是否已达100步另保留未知，不能将本地测试达标当作交接成功。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 6.4,
+    "log_step": 344,
+    "evidence_zh": "观察：视频5.8–7.8 s及视频10.0–18.0 s，白色假肢手保持张开，伸展手指位于肌肉右手的下方，红边盒体则停在右腕上侧的起始柱缘，没有进入白手可包夹的位置，也没有随白手移动。推断：接收手与盒体的相对位置未形成有效承接；不据此断言假肢从未发生过轻触。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "观察：log step 344在整段评分动作中将假肢手10个手指执行器目标设为-1.0并维持固定交接臂姿；log step 354的对应脚本明确将此设定注释为张开、伸展手指。视频约6.4–18.0 s显示白手伸指停在右手下方，物体仍依靠起始柱缘与右腕，没有被白手承载。推断：实际接收姿态与指形未建立能接走物体的承载或对向夹持布局；现有证据未验证本次相同状态下的替代动作。"
+   }
+  ],
+  "gap_zh": "没有状态日志，也没有评分局逐步接触标签或两臂累计接触计数；白手与盒体的微小接触界面被右腕及起始柱遮挡，因此假肢接触100步节点无法定值。log step 193、267、273的接触计数是本地测试，log step 345与346证实本地和远端关节状态存在差异，不能转用其计数或达标时刻；肌肉臂节点达标时间为视觉估计。评分器未报告DAG所列五项原生指标，均记null。",
+  "unmet": {
+   "object_held_by_myo": true,
+   "object_handed_to_prosthesis": null,
+   "object_on_goal_pillar": false
+  }
+ },
+ "summary_zh": "评分局中，肌肉右手接触并将盒体沿起始柱面移至柱缘，按累计接触的实际验收估计完成右臂节点；白手靠近后没有携物承接，目标柱放置未完成。断点在交接操作：接收手保持张开并停在右手下方，未形成有效承载或夹持布局。假肢累计接触是否达到100步因界面遮挡及缺少评分局接触历史而未知，agent报告的415/128步来自本地测试，不能用于评分局计数。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "抽查视频0.5–18 s各帧：盒体始终在起始柱上，右手肌肉臂接触，白手伸到右手下方未携物，未上目标柱；节点值、断点（交接）、位姿不准与抓取和接触设计根因与证据一致，假肢节点保留未知合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:45:43Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--d2hna7y/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--d2hna7y/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0ef4f239ba737e2c7bb6d42e9785d169ee873f4508ff7fa7c1579fbe4d170951",
+   "video_sha256": "8922835744863e63ef6805b2c29bef1c8197f7423a5be4ef51c1b8bb6bc9860a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--fyvhoer.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--fyvhoer.json
new file mode 100644
index 0000000000000000000000000000000000000000..8326480a0ef00a6f363b65610f086f3f831494b9
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--fyvhoer.json
@@ -0,0 +1,259 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-standard--fyvhoer",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 1000,
+   "sim_steps": 1000,
+   "physics_steps": 5000,
+   "steps": 1078,
+   "calls": 5,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 1078,
+   "elapsed_s": 3170.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.306,
+   "peak_processes": 12,
+   "peak_disk_gb": 0.601,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-standard--fyvhoer/replay.mp4",
+  "duration_s": 40.04,
+  "fps": 25.0,
+  "frames": 1001,
+  "sha256": "c042cf5f258790fa973b7eb0c38d184a1621a8acbfd1f2be86bd2bd856f744d4"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [
+     [
+      4.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频约0.5–5.5 s的逐0.25 s双画面中，右侧红色肌肉手的弯曲手指及掌部贴压红边长方体，物体倾倒后仍由起始柱支承；视频5.75–6.5 s右手撤走，物体留在原柱，未见持物抬升。"
+     },
+     {
+      "t": 4.5,
+      "log_step": 65,
+      "text_zh": "推断：log step 65提交的是最后第3局1000个控制步，视频1001帧、25 fps，故每控制步约对应视频0.04 s；脚本在控制步140切换肌肉命令，与视频约5.6 s右手开始撤离相符。可见贴压从视频约0.5 s持续至5.5 s，累计100步约在视频4.5 s达到，时间误差约0.5 s；按本DAG实际累计接触验收记达成并锁存，而不额外要求真正抬起。此判断不把本地优化输出的116步当成本局计数。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频约0.5–5.5 s，右侧肌肉手的手指及掌部压碰起始柱上的红边长方体；物体受压倾倒但没有被合手夹住并带离柱面。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.0,
+      "log_step": 65,
+      "text_zh": "观察：视频5.75–6.5 s，肌肉右手空手撤离，红边长方体仍在起始柱；视频0–40.0 s的全片粗查及视频39.04–40.0 s末段细查，白色假肢手始终远离起始物体，没有双手会合、承接或携物移动。log step 65最终提交代码将假肢17个执行器在全部1000步均设为初始关节位置的保持命令，只切换右臂肌肉命令。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 65,
+      "text_zh": "推断：右手离开前物体未被取走，此后两手无交接动作，且最终提交脚本无假肢接收阶段，因此假肢累计接触条件不成立；并非仅由总成功分数0推断。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.5,
+      "log_step": 65,
+      "text_zh": "观察：视频6.0–6.5 s，红边物体仍留在起始柱，肌肉手已空手离开；后续至视频40.0 s，两手没有携物到另一根柱顶再松开的动作。成功参考视频约17–25 s可见两手靠拢及向另一柱运送的动作，本局没有相应过程；参考只用于辨认可见线索。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 65,
+      "text_zh": "推断：本局未发生离开起始柱后的搬运，假肢手全程执行固定保持命令，故物体放上独立目标柱及柱独占接触尾窗均未达成；视频约7 s后原柱离开手部相机视野，不声称末帧直接显示物体位置。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_handed_to_prosthesis",
+   "t": 5.7,
+   "log_step": 65,
+   "where": "首步前",
+   "evidence_zh": "观察：视频约5.7 s肌肉右手撤走而物体留在起始柱，白色假肢手没有迎向或接触物体，此后至视频40.0 s也未开始交接。log step 65最终第3局仅提交右手触碰后撤离及假肢定姿保持命令；此前log step 64的第1局在78步结束后已被重置，不属于本视频。断点为未启动且未补做的交接步骤，视频5.7 s仅为右手撤离的导航锚点。"
+  },
+  "unmet": {
+   "object_held_by_myo": true,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 5.7,
+    "log_step": 65,
+    "evidence_zh": "观察：视频约0.5–5.5 s只有肌肉右手在起始柱处触碰物体，视频5.75–6.5 s空手撤离；白手在视频全程及39.04–40.0 s末段都未伸向物体，未启动交接。log step 65的最终提交中，假肢所有执行器始终保持初始关节目标，没有接收动作。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：会话log step 44、50的本地肌肉抓持搬运测试反复输出post=0，log step 62–63的完整本地控制器测试仍输出PROSTH_SHORT(0)、NO_GOAL，物体没有被送到假肢接收处；这些是控制开发的会话证据，不是本视频的状态计数。其后log step 65明确换成只触碰物体、假肢定姿保持的最后一局，与视频5.75–6.5 s空手撤离相符，表明未形成可执行的交接控制而退回触碰基线。现有证据不足以进一步区分抓取几何、逆解或跟踪问题；不将用时、先前局的掉落或agent关于工作空间不可达的自述当作本局物理现象。"
+   }
+  ],
+  "gap_zh": "无状态日志，评分器字段未报告本局两臂精确接触计数、goal_dist、goal_touch或末尾柱独占步数；右臂100步达标及视频约4.5 s时刻由持续可见接触和每控制步对应一视频帧推断，可能相差约0.5 s。视频约7 s后起始柱离开两路手部相机视野，末帧不能直接核实物体位置，但此前物体留柱、全程无交接动作及最终固定假肢命令足以判定后两节点未达。日志中的本地副本测试与被重置的第1局仅作控制开发背景，不用于给本局计数；控制根因无法细分，agent自述假肢工作空间限制也未独立证实。"
+ },
+ "summary_zh": "受评的是log step 65最后重置后的第3局：右手在视频约0.5–5.5 s贴压起始柱上的物体，按累计接触100步的实际验收，第一节点推断于视频约4.5 s达成，虽未真正持起，最终进度为1/3。视频约5.7 s右手空手撤离，假肢手未启动交接，后续放上目标柱也未达；会话控制测试未形成可用交接控制，最后退回了只触碰的基线。无本局精确接触计数，且视频约7 s后原柱不在视野内，因此第一节点时刻与控制细分原因仍有不确定性。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "标签与视频、log一致：右手视频约0.5–5.5 s持续贴压起始柱上的物体（约125控制步，足以支持100步累计的推断），白手全程未动，后两节点未达，断点首步前+没尝试+控制（子类不明）均有证据支持，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:41:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--fyvhoer/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--fyvhoer/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d15636894ab7542080ae32f6bcb2ba5e47834a5ffe5bde78a232f5176e65b520",
+   "video_sha256": "c042cf5f258790fa973b7eb0c38d184a1621a8acbfd1f2be86bd2bd856f744d4",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--mnb62cw.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--mnb62cw.json
new file mode 100644
index 0000000000000000000000000000000000000000..c2c91f59040f48a6a4f077af2acc185d7166fa32
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--mnb62cw.json
@@ -0,0 +1,254 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-standard--mnb62cw",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 739,
+   "sim_steps": 739,
+   "physics_steps": 3695,
+   "steps": 6182,
+   "calls": 2321,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 6182,
+   "elapsed_s": 5362.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-standard--mnb62cw/replay.mp4",
+  "duration_s": 29.6,
+  "fps": 25.0,
+  "frames": 740,
+  "sha256": "05941e6d344104932f5ddd781b41841fe3d1c0eb856f11b7244bdcf61e2ad826"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [
+     [
+      9.8,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 5.8,
+      "log_step": null,
+      "text_zh": "观察：视频约5.8–11.6 s，肌肉右手拇指持续抵住直立物体的上缘，物体仍由起始圆柱承托；视频11.8–13.6 s，其他手指进一步收拢到物体上缘两侧。"
+     },
+     {
+      "t": 9.8,
+      "log_step": 231,
+      "text_zh": "推断：本节点依DAG实际验收口径只需MYO累计接触100个控制步，不另要求抬升。该局739个控制步对应740帧、25 fps，视频时间约为control_steps/25；log step 231的334步对应视频13.36 s，与视频13.4 s的收指结束、开始抬手核对一致。视频约5.8 s起的持续触碰到视频约9.8 s已覆盖约100步，故以视频9.8 s为近似达标锚点（约±0.5 s），事件随后保留；这不表示已抓稳或带离起始柱。"
+     },
+     {
+      "t": 14.2,
+      "log_step": 236,
+      "text_zh": "观察：视频13.4–15.4 s，肌肉右手抬起时物体从上缘夹持处转落，平躺回起始柱；视频18.4–23.2 s的再次收指、抬手也未带走物体。log step 236中agent自述“The lift failed again, and the box is now flat on the pillar”，与该局视频一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "肌肉右手先以拇指抵压物体上缘，随后拇指与其余手指在物体上缘两侧收拢作边缘夹持并尝试抬升；夹持未能稳定承重，物体回落到起始柱。末段右手再次收指并拨动物体，使其离开柱缘。仅右手与物体发生此接触，白色假肢手未共同夹持。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 17.2,
+      "log_step": 235,
+      "text_zh": "观察：视频16.0–24.0 s的连续抽帧中，白色假肢手向身体附近的接收区抬起，但与起始柱上的红边物体始终分离；肌肉右手仍在起始柱处反复收指重抓。log step 235仅令假肢手移向接收姿态，没有已完成交接的服务读数。"
+     },
+     {
+      "t": 29.56,
+      "log_step": 242,
+      "text_zh": "观察：视频24.8–29.56 s，白手仍远离物体；物体先留在起始柱，再从柱缘掉落，未随白手移动。没有成功参考视频约17–20 s中两手在物体处会合并由白手持物的可见过程；本局未开始实际的物体交接，假肢臂累计100步物体接触的条件不成立。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 15.4,
+      "log_step": null,
+      "text_zh": "观察：视频0–27.6 s，物体一直在肌肉右手旁的起始圆柱上，被触碰、夹住上缘并翻倒；没有被运往另一目标柱。"
+     },
+     {
+      "t": 29.56,
+      "log_step": 242,
+      "text_zh": "观察：视频27.8–29.56 s，物体越过起始柱边缘并向下掉落，末帧肌肉手为空，白手没有持物，也没有物体留在目标柱上的画面。log step 242的服务状态为episode 11、control_steps 739、done True、reason为“failure: object dropped or the time limit passed”；评分记录task_failure=1、timeout=0，与视频掉落一致。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_held_by_myo",
+   "t": 14.2,
+   "log_step": 232,
+   "where": "操作中",
+   "evidence_zh": "观察：视频13.4–15.4 s，肌肉右手正进行收指后的抬升检验，物体却从上缘夹持处转落并平躺回起始柱；log step 230的脚本在收指后命令上移，log step 232仍在等待这次抬升结果。视频18.4–23.2 s及24.8–29.56 s的重抓均未修复承重夹持，最后物体跌出柱缘。断点按实际正在操作的右手拿取步骤记录；该节点较弱的累计触碰验收已达标，并不等于物理抓取已成功，也不能据此把断点移到尚未开始的交接或目标柱放置。"
+  },
+  "unmet": {
+   "object_held_by_myo": true,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 14.2,
+    "log_step": 236,
+    "evidence_zh": "观察：视频11.8–13.6 s，肌肉右手拇指与其余手指从两侧夹住物体上缘；视频13.8–15.4 s抬手时，物体相对手指转动、滑脱，仍落在起始柱面上，未随手移动到接收区。视频27.8–29.56 s的最后一次抓取后物体又跌出柱缘，白手未承接；log step 236中agent自述抬升失败，与前一段画面一致。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 13.4,
+    "log_step": 230,
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频11.8–14.4 s显示肌肉右手主要捏住长方体上缘的小片区域，没有形成能把整个物体随手带走的承载夹持；手上移时物体在夹持处转动并脱离。log step 230的脚本在收指后直接安排抬升，log step 236、241中agent分别自述抬升失败和只勾到边缘，与视频18.4–23.2 s的边缘重抓、空手抬起相符。证据支持接触布局不能稳定承重；未在本局验证一套可行替代握法，不能进一步断定具体肌肉参数或接触力数值。"
+   }
+  ],
+  "gap_zh": "未提供状态日志或逐控制步接触标签，评分字段也未报告五项原生指标；右臂100步累计接触的达标时刻由可见持续触碰及逐控制步渲染映射估计，无法精确恢复接触计数、力或三维距离。右手接触达标与实际拿取失败并存，源于DAG已说明的验收口径差异；没有证据把本次物理滑脱归因于任务设计。"
+ },
+ "summary_zh": "第11局的肌肉右手持续触碰物体，按DAG累计接触100个控制步的实际口径，右手节点约在视频9.8 s达标并保留，最终进度为1/3；这不代表物体已被稳定持起。实际操作在视频约14.2 s的右手抬升中出现边缘夹持滑脱，之后重抓未修复，物体在视频27.8–29.56 s跌出起始柱，未交给白手或放上目标柱，根因为抓取和接触设计。无状态日志和接触计数，达标时刻及接触布局的解释含视觉推断，原生指标均未报告。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点值、断点、现象与根因与视频一致：右手约5 s起触碰物体、13.4 s后抬升时滑脱、末段物体跌出柱缘，白手始终未接触物体。节点达标时刻属DAG口径下的合理推断，已标为推断。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:43:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--mnb62cw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--mnb62cw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "04ce3f1cd0b995c421b1e1439b3d0bb510fb992852a2893dbb693ba133a34bb9",
+   "video_sha256": "05941e6d344104932f5ddd781b41841fe3d1c0eb856f11b7244bdcf61e2ad826",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--x4egqsj.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--x4egqsj.json
new file mode 100644
index 0000000000000000000000000000000000000000..8b5c5aee63fb9e16291fdb1952aedad428d853ff
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--x4egqsj.json
@@ -0,0 +1,260 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-standard--x4egqsj",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 1000,
+   "sim_steps": 1000,
+   "physics_steps": 5000,
+   "steps": 3400,
+   "calls": 79,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 3400,
+   "elapsed_s": 1609.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.706,
+   "peak_processes": 14,
+   "peak_disk_gb": 1.548,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-standard--x4egqsj/replay.mp4",
+  "duration_s": 40.04,
+  "fps": 25.0,
+  "frames": 1001,
+  "sha256": "12b02f935fc7bc59b6cee67f5d363b9657fdd9ee80280e5d0a2f57c756e4972b"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.0,
+      "log_step": 182,
+      "text_zh": "观察：视频 0.0–2.0 s 的双画面中，红色肌肉右手张开并从红边灰盒旁向下离开，盒子仍立在起始柱顶，没有随手抬升或被包夹。log step 182 的最后一次重置后先提交200步恒定肌肉命令；该步后处理报错，但已提交的运动属于第11局。"
+     },
+     {
+      "t": 33.2,
+      "log_step": 192,
+      "text_zh": "观察：视频 30.4–38.4 s 的接近段中，右手在盒子下方或侧方摆动；视频 33.2、34.8、35.6、36.8、38.4 s 可见盒子仍在原起始柱上，右手未形成包夹，未见持续的物体接触。视频 39.0–40.0 s 双手仍空着。"
+     },
+     {
+      "t": 30.4,
+      "log_step": 192,
+      "text_zh": "推断：本局1000个控制步对应1001帧、25 fps，视频每步约0.04 s；log step 192 从本局control_steps=760开始接近，对应视频 30.4 s，已用视频 30.4–30.8 s 手臂从保持转为摆动核对。全局未见可累计达100步的右臂接触，故按DAG的历史接触验收判未达，而非仅因没有抬高物体判未达。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "候选名“未建立接触的接近尝试”：视频 30.4–38.4 s 肌肉右手手指向起始柱上的红边灰盒接近，但未见夹住或推动盒子；不能标为实际抓握、推/扫或压/擦。短暂擦碰无法从画面排除，未虚构接触计数。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 35.6,
+      "log_step": null,
+      "text_zh": "观察：视频 25.0–30.4 s 以及视频 33.2、35.6、36.8、38.4 s，盒子仍在起始柱，白色假肢手与盒子分离；视频 39.0–40.0 s 白手仍空着。全片未见物体随白手运动，亦未见足量的白臂接触。成功参考录像的视频 20.0 s 则能看见白手包住并承接盒子，此处仅用于辨认可见线索。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 192,
+      "text_zh": "推断：本局末段脚本（log step 192）仍在对右手做接近，未进入交接控制；结合盒子始终留在原柱的可见片段与白手空手的后续画面，假肢臂累计100步接触节点未达。没有把被重置的早期试验算入本局。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 38.4,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0、1.4、22.0、25.0–30.4、33.2、35.6、36.8、38.4 s 盒子保持在同一起始柱上，未见被搬往另一根目标柱；视频 39.0–40.0 s 两手仍空着，没有放置或松手动作。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 197,
+      "text_zh": "推断：结合全片未出现搬运、可见片段中盒子仍在起始柱，以及 log step 197 服务在本局1000个控制步给出failure，物体放上目标柱节点未达；该判断不是仅从success=0倒推。末帧盒子在画外，不据此声称直接观察了末帧柱面关系。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_held_by_myo",
+   "t": 33.2,
+   "log_step": 192,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 30.4–38.4 s 机器人肌肉右手尝试接近起始柱上的盒子，却在其下方或侧方摆动；视频 33.2 s 盒子仍在柱上，未被夹住，后续也未修复为持物。log step 192 正在给右手求逆解并分块提交接近命令；log step 197 本局1000步后服务结束并判失败。断在右手持起这一步，后两步尚未开始。"
+  },
+  "unmet": {
+   "object_held_by_myo": false,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 33.2,
+    "log_step": 192,
+    "evidence_zh": "观察：视频 30.4–38.4 s 机器人右手向盒子附近摆动却没有包夹、承载盒子；视频 33.2、35.6、38.4 s 盒子仍在起始柱，视频 39.0–40.0 s 右手仍空着。所需的持物接触未建立，未见先持有再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 188,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：右臂的命令执行存在明显偏差，使手未能到达并保持在物体接触位姿。log step 188 的脚本要求保持当前关节姿态，提交120步后打印的关节绝对误差中包含0.979和1.527 rad；log step 192 明确以盒子附近指尖位置为目标求逆解，结果指尖仍偏离目标，视频 30.4–38.4 s 亦见空手反复摆动。证据支持控制层问题，但未验证具体接触方案或哪组参数能成功，故不细分为参数调试、反馈控制或抓取设计。"
+   }
+  ],
+  "gap_zh": "没有状态日志，也没有本局物体接触标签、两臂历史接触累计数、goal_dist、goal_touch和末尾独占接触计数；native_metrics均未报告。随手移动的相机在部分时段及视频 39.0–40.0 s 将盒子遮挡或移出画面，最终未达依据此前反复可见的起始柱关系、空手动作和服务终局结果综合判断。微小擦碰不能精确排除，但无达标的持续接触或搬运迹象；控制失配的具体机制未被查明，log step 182 的后处理索引错误发生在动作提交后且本局继续运行，不能把它当作外部中断。"
+ },
+ "summary_zh": "第11局为受评局：物体始终未被从起始柱带走，右手持起、交给左手和放上目标柱三个节点均未达。视频 30.4–38.4 s 的右手接近未建立持物接触，日志中的姿态保持误差支持控制层问题，最终本局1000个控制步后判失败。缺少接触历史和状态日志，短暂擦碰及控制失配的具体机制不能精确确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "三个节点均未达、断点在右手持起、现象没建立接触、根因停在控制，与视频及服务终局读数一致；未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:43:12Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--x4egqsj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--x4egqsj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c2da55792d751ffaab6a2916a1b04f456140ef5eec07876a12f550f7673955f8",
+   "video_sha256": "12b02f935fc7bc59b6cee67f5d363b9657fdd9ee80280e5d0a2f57c756e4972b",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--ydvnvvj.json b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--ydvnvvj.json
new file mode 100644
index 0000000000000000000000000000000000000000..d7da01dc2526203d9d2a6b61a9ade0530787f9be
--- /dev/null
+++ b/phase2/myosuite/bimanual/trajectories/myosuite-bimanual-i00-standard--ydvnvvj.json
@@ -0,0 +1,244 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-bimanual-i00-standard--ydvnvvj",
+ "taskKey": "myosuite/bimanual",
+ "dag_sha256": "0af83fb75572a8dee8e6c0b8abc848886eb871d14f1068bf9e0ea0f234e3bb96",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 1,
+   "sim_steps": 1,
+   "physics_steps": 5,
+   "steps": 1,
+   "calls": 2,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 1,
+   "elapsed_s": 3640.0,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.366,
+   "peak_processes": 7,
+   "peak_disk_gb": 0.474,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-bimanual-i00-standard--ydvnvvj/replay.mp4",
+  "duration_s": 0.08,
+  "fps": 25.0,
+  "frames": 2,
+  "sha256": "5114182e3c768f781d467ac2efc0f338c13a41aba77051463293636e57482924"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_held_by_myo": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：已查看本次视频全部两帧（视频 0.00 s、0.04 s）；右侧视图中红边长方体仍立在起始圆柱上，带红色肌肉束的手张开停在其上方，手指与物体之间可见间隙，没有靠近抓取或持起动作。"
+     },
+     {
+      "t": null,
+      "log_step": 57,
+      "text_zh": "观察：log step 57 中 agent 通过实时机器人服务 r.submit 仅提交一行80维全零执行器指令，dt=0.01；服务回报 control_steps=1。run.json 的评分字段也记录 n_actions=1、sim_steps=1、resets=0。"
+     },
+     {
+      "t": null,
+      "log_step": 57,
+      "text_zh": "推断：受评局总共只有一个控制步，不可能达到本节点要求的肌肉臂累计接触至少100个控制步；此结论同时与视频中未抓取的画面一致，并非仅根据失败评分反推。"
+     }
+    ]
+   },
+   "object_handed_to_prosthesis": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.04,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s、0.04 s 的两个视图均没有两手在长方体处会合、白色假肢手承接或物体随假肢手移动；物体始终留在起始圆柱上。"
+     },
+     {
+      "t": null,
+      "log_step": 57,
+      "text_zh": "推断：log step 57 的实时服务仅执行一个控制步，且评分字段 sim_steps=1、resets=0，因此受评局也不可能达到假肢臂累计接触至少100个控制步。log step 159 的承接记录来自 agent 另建的本地仿真测试，不能算作受评交接。"
+     }
+    ]
+   },
+   "object_on_goal_pillar": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.04,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s、0.04 s 中红边长方体一直在起始圆柱上，没有离开原位或被送至另一个目标柱，两个视图均未出现放置动作。"
+     },
+     {
+      "t": null,
+      "log_step": 159,
+      "text_zh": "观察：log step 159 的 solved=True、goal_dist=0.0696 与 goal_touch=11 是本地测试脚本的输出；该脚本在 log step 156 中以 gym.make 新建环境并调用 env.step，未调用实时服务 r.submit。这些数值不属于本次受评局，故不用于节点达成或原生指标。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "goal_dist",
+   "unit": "m",
+   "threshold": 0.17,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "myo_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "prosthesis_contact_count",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_touch",
+   "unit": "control_steps",
+   "threshold": 10,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "goal_only_tail_count",
+   "unit": "control_steps",
+   "threshold": 8,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_held_by_myo",
+   "t": 0.0,
+   "log_step": 57,
+   "where": "首步前",
+   "evidence_zh": "观察：本次视频全部两帧（视频 0.00 s、0.04 s）中物体仍在起始柱上，肌肉手未抓取物体。log step 57 是唯一实时动作提交，只执行一个全零指令控制步；评分记录 n_actions=1、sim_steps=1，受评局未启动右手持起物体这一步。"
+  },
+  "unmet": {
+   "object_held_by_myo": false,
+   "object_handed_to_prosthesis": false,
+   "object_on_goal_pillar": false
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 57,
+    "evidence_zh": "观察：视频 0.00 s、0.04 s 中肌肉手保持张开并与长方体分离，长方体仍在起始柱上；log step 57 仅由实时服务执行一行全零诊断指令。后续 log step 156–160 的动作在 agent 本地 gym 环境测试和修改，受评录像没有抓取尝试。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "观察：inputs/ 没有状态日志；本次录像只有视频 0.00 s、0.04 s 两帧，评分字段未报告各项原生指标的实测值，故这些指标均留空。log step 159 虽显示本地测试通过接触历史检查，log step 160 仍仅修改该本地脚本；完整记录没有说明为何未将后续控制序列用于实时服务，也没有最终完成声明。推断：可以确定受评抓取未启动，但不足以把原因细分为规划、控制或交付操作错误；不能把超时或缺少后续提交本身当作根因，因此根因层留空。"
+ },
+ "summary_zh": "受评录像中物体始终留在起始柱上，右手持起、交给左手、放上目标柱三个节点均未达成（0/3）；断点为右手持起物体的首步前，现象为没尝试。log step 57 仅向实时服务提交一个全零控制步，而 log step 159 的成功接触与放置结果来自独立本地仿真，未出现在受评录像中。缺少未执行后续实时动作的决策证据与状态日志，根因留空，原生指标不采用本地测试值。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标注与证据一致：受评局仅 1 个控制步（log step 57、run.json n_actions=1），视频两帧物体仍在起始柱上，三个节点均未达成；断点首步前 + 没尝试，根因层留空并写 gap_zh，符合 FAIL-22 与 D23；log 159 的本地测试结果未被当作受评事实。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:42:53Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--ydvnvvj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-bimanual-i00-standard--ydvnvvj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b915383042d8ea3799f7d581e92e19aabe2e66e5ce002ed7ab37d02cf6768976",
+   "video_sha256": "5114182e3c768f781d467ac2efc0f338c13a41aba77051463293636e57482924",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--346fbz4.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--346fbz4.json
new file mode 100644
index 0000000000000000000000000000000000000000..a04caf6c3650753c6e361e2c6c400b3c11084ee7
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--346fbz4.json
@@ -0,0 +1,183 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-privileged--346fbz4",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 172,
+   "sim_steps": 172,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 103,
+   "agent_resets": 12396,
+   "agent_sim_steps": 11699,
+   "agent_records": 3,
+   "agent_replays": 9,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.274,
+   "peak_processes": 16,
+   "peak_disk_gb": 0.241,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-privileged--346fbz4/replay.mp4",
+  "duration_s": 6.92,
+  "fps": 25.0,
+  "frames": 173,
+  "sha256": "d13fe45044a18ee63b0e666a50055dab5d7fbbe19f8a5f0d92facb04caaa4cfc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 白色 MyoLegs 模型位于场地中部，浅蓝圆顶黑色柱状对手位于画面右下，两者明显分离；视频 0.0–5.6 s 模型摆腿并向对手方向移动，但尚未靠近对手。"
+     },
+     {
+      "t": 6.88,
+      "log_step": 129,
+      "text_zh": "观察：视频 5.6–6.88 s 的密集帧显示模型身体下沉并向前倒伏，末帧骨盆与对手仍有明显水平间隔，对手圆顶仍为浅蓝色；成功参考视频 9.64 s 则显示模型骨盆紧邻对手、圆顶变绿。log step 129 对最终交付文件的两次重放均报告 success=false、172 步、相同 state_sha，与评分器 replay_success=0、deterministic=1 一致。"
+     },
+     {
+      "t": 6.88,
+      "log_step": 134,
+      "text_zh": "推断：结合全段视频未出现追及位置关系及原生重放失败结果，追及对手节点始终未达成。log step 134 再次验证交付文件为 172 步且失败；log step 132 的另一组参数测试及 log step 139 对更近样本的自述不代表本次交付视频。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0.0–5.6 s，MyoLegs 双腿摆动、脚部先后落在木质地面承重，推动本体向对手移动；视频约 6.0 s 后支撑失稳、身体下沉。未见本体与柱状对手实际触碰，任务验收也不要求触碰。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 6.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 6.0 s，正在向固定对手行进的 MyoLegs 身体开始明显前倾下沉，视频 6.4–6.88 s 持续倒伏，未恢复支撑；回合结束时仍与对手分离。log step 129 对已交付的 172 步文件确认两次重放均失败，评分器 task_failure=1。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 6.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 5.6–6.88 s，MyoLegs 从尚能以腿支撑的姿态转为躯干前倾、骨盆明显降低及身体倒伏，直到末帧未恢复；模型此时仍在场内且尚未追及对手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent 的步态控制器未能在向对手推进时维持身体支撑。log step 127 用 best4 参数录制并保存 172 步动作，原生 replay 即报告失败；log step 134 与 log step 129 对该交付重复验证仍失败，视频约 6.0–6.88 s 显示其失稳倒伏。agent 在 log step 139 自述支撑腿近似直杆、未能制动前向运动，这是尚未独立验证的具体机制；因此根因仅记控制，不细分，也不将已修复的早期求解器问题作为本次根因。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志；俯视视频分辨率有限，不能精确给出骨盆高度越过 0.5 米阈值的时刻、逐足接触力或控制失稳的细分类机制。评分器 reward 未报告 distance，故原生距离指标留 null；log step 124 的距离为会话测试输出，不作为评分器指标。视频时钟独立标注，未把 agent 自述的仿真持续时间映射到视频。"
+ },
+ "summary_zh": "本次交付的 MyoLegs 已摆腿向固定对手移动，但始终未达到追及距离，最终完成 0/1 个节点。视频约 6.0 s 起模型失去支撑并倒伏，断点为追及对手的操作中，现象为全身·平衡；日志对同一交付文件的确定性失败重放支持控制未能维持支撑的判断。缺少逐帧状态和接触力，不能确定更细的控制机制，评分器也未报告原生距离数值。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "节点未达成、断点（追及对手·操作中，约6.0 s开始倒伏）、现象全身·平衡、根因停在控制均有视频与重放日志支持；视频全程模型与对手明显分离，未见错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:03:29Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--346fbz4/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--346fbz4/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "43d715927c77ecb185bce10520f9165eac3e8d9facc677bde44d387d8ff95be5",
+   "video_sha256": "d13fe45044a18ee63b0e666a50055dab5d7fbbe19f8a5f0d92facb04caaa4cfc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--bgpmfjx.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--bgpmfjx.json
new file mode 100644
index 0000000000000000000000000000000000000000..3535d0534f4212f51389e08780c47c198d1e377c
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--bgpmfjx.json
@@ -0,0 +1,159 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-privileged--bgpmfjx",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 242,
+   "sim_steps": 241,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 161,
+   "agent_resets": 164,
+   "agent_sim_steps": 15557,
+   "agent_records": 0,
+   "agent_replays": 38,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.282,
+   "peak_processes": 20,
+   "peak_disk_gb": 0.983,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-privileged--bgpmfjx/replay.mp4",
+  "duration_s": 9.68,
+  "fps": 25.0,
+  "frames": 242,
+  "sha256": "4bb26d8f44f912cf4fab08f2272dcaea35e8d29f64707e3f5bed21275d86eb13"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [
+     [
+      9.64,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s，白色MyoLegs模型位于场地中央偏远处，带浅蓝圆顶的黑色柱状对手位于其右前方，两者明显分离；视频 0.00–8.08 s，模型用双腿迈步并逐渐靠近固定对手。"
+     },
+     {
+      "t": 9.64,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 9.20–9.64 s，白色模型继续靠近对手，视频 9.60 s 对手顶部仍为浅蓝色，视频 9.64 s 末帧变为绿色，模型已到其近旁。该时间取自视频帧时间戳，未将 agent 工作时间或模拟时间换算为视频时间。"
+     },
+     {
+      "t": 9.64,
+      "log_step": 289,
+      "text_zh": "推断：视频 9.64 s 的变绿与最终成功判定一致，追及对手节点在此达成。inputs/run.json 的评分器字段 success=1、replay_success=1、deterministic=1、instance_ok=1、task_failure=0；log step 250 安装的242行交付文件，在 log step 289 的最终官方重放中两次均于控制步241返回 success=true 且最终状态哈希相同。更晚另存的256行测试候选不属于该交付文件；未用会话测试中的跌倒或距离替代本段视频。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0.00–8.08 s，MyoLegs的双脚交替落在木质场地上支承身体，双腿迈步使骨盆向固定对手移动；实际可见的支承交互为脚与地面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "白色MyoLegs模型从场地中央迈步靠近固定对手，视频 9.64 s 末帧对手顶部变绿；结合评分器成功字段及 log step 289 的最终交付重放，追及对手节点完成，终态为1/1，无失败断点。没有逐帧状态日志，评分器未报告原生distance数值，故不从低分辨率俯视画面估算精确距离，达成时刻以视频变色帧为锚。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功回合，追及对手节点在视频 9.64 s 末帧（对手顶部由浅蓝变绿）达成，与评分器 success=1 及 log step 250/289 的官方重放（控制步241）一致；contact_actual 为脚支撑，无失败断点，未发现需修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:01:42Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--bgpmfjx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--bgpmfjx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9dcb53e86afd7b0e35180d60c2ae1965294f47b4c2c13c4c6fca06fce83b9c0c",
+   "video_sha256": "4bb26d8f44f912cf4fab08f2272dcaea35e8d29f64707e3f5bed21275d86eb13",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--dc2fmax.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--dc2fmax.json
new file mode 100644
index 0000000000000000000000000000000000000000..2cb3bc2070e1d4f6eb994c1423d0dbcadf30e320
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--dc2fmax.json
@@ -0,0 +1,162 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-privileged--dc2fmax",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 260,
+   "sim_steps": 260,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 46,
+   "agent_resets": 8362,
+   "agent_sim_steps": 72813,
+   "agent_records": 5,
+   "agent_replays": 10,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-privileged--dc2fmax/replay.mp4",
+  "duration_s": 10.44,
+  "fps": 25.0,
+  "frames": 261,
+  "sha256": "701c0673438de02a08f9998134753b36166537c2bc647093662e4dfb5a3fea32"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [
+     [
+      10.4,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s，白色 MyoLegs 模型位于场地中部，带浅蓝圆顶的黑色柱状对手在其右前方，两者明显分离；视频 0.00–9.20 s，模型通过双腿迈步逐渐向固定对手接近。"
+     },
+     {
+      "t": 10.4,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 9.20–10.40 s，模型继续靠近对手；视频 10.36 s 圆顶仍为浅蓝色，视频 10.40 s 末帧圆顶变为绿色。成功参考视频的末帧（参考视频 9.64 s）也出现同样的绿色信号；这里只用参考视频解释画面信号，本次完成时刻取自本次视频。"
+     },
+     {
+      "t": 10.4,
+      "log_step": 146,
+      "text_zh": "观察：log step 143 中，最终交付动作的两次新进程重放均报告 success=true、steps=260、state_sha=87fb24c9797074df；log step 146 对该最终文件的渲染重放报告 solved=1、reason=success，哈希相同。inputs/run.json 的评分器字段也为 success=1、replay_success=1、deterministic=1、instance_ok=1，与视频 10.40 s 的绿色信号一致。"
+     },
+     {
+      "t": 10.4,
+      "log_step": null,
+      "text_zh": "推断：结合视频 10.40 s 的达标信号和评分器成功结果，“追及对手”在视频 10.40 s 达到骨盆水平距离不大于 0.5 m 的原生成功条件，并保持至录像结束；本次没有逐帧状态日志，不从低分辨率画面测量精确米数。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0.00、2.00、4.00、6.00、8.00 s 及其间帧中，MyoLegs 的双腿屈伸迈步，脚先后在木质地面落地承重，使身体向对手移动；实际作用端为双脚，接触对象为地面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "MyoLegs 通过双腿迈步接近固定对手，在视频 10.40 s 末帧触发圆顶变绿的成功信号，完成“追及对手”（最终 1/1）；评分器确认成功且最终交付可确定性重放。没有逐帧状态日志，精确距离无法由画面测量，且评分器 reward 未报告 distance 数值，因此该原生指标留空；log step 144 的仿真 2.60 s 为 agent 脚本诊断时间，不作为视频时间。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "节点“追及对手”在视频 10.40 s 末帧圆顶变绿，与评分器成功字段和 log step 143/146 的重放结果一致；时间、接触（腿/躯干·支撑）和 failure=null 均成立。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:05:08Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--dc2fmax/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--dc2fmax/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "acc1a0406e214b3f2dde0752e2142c183a24d39d657f36533fce0b9613708190",
+   "video_sha256": "701c0673438de02a08f9998134753b36166537c2bc647093662e4dfb5a3fea32",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--ddgtnpd.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--ddgtnpd.json
new file mode 100644
index 0000000000000000000000000000000000000000..3f10577ede5f0f323ca392cdea4df3d322fef204
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--ddgtnpd.json
@@ -0,0 +1,183 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-privileged--ddgtnpd",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 61,
+   "sim_steps": 61,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 29,
+   "agent_resets": 313,
+   "agent_sim_steps": 17046,
+   "agent_records": 6,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.659,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.239,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-privileged--ddgtnpd/replay.mp4",
+  "duration_s": 2.48,
+  "fps": 25.0,
+  "frames": 62,
+  "sha256": "5c8ab98b7c02335f67c3e96aa1368a17ecc4b42930d155e1a5b19e457527d1df"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00 s中，白色MyoLegs模型位于场地中后部，带浅蓝圆顶的黑色柱状对手位于画面右前方，两者明显分离。"
+     },
+     {
+      "t": 2.44,
+      "log_step": null,
+      "text_zh": "观察：检查视频0.00–2.44 s全程及视频1.28–2.44 s逐帧末段，MyoLegs仅向前移动少许便逐渐倾倒，直到末帧骨盆与对手仍明显分离，从未出现追及所需的近距离位置关系。"
+     },
+     {
+      "t": null,
+      "log_step": 42,
+      "text_zh": "观察：log step 42中，agent脚本重新生成并保存61×80动作的交付文件；基准CLI对该文件两次重放均返回success=false、steps=61，末态哈希相同且instance_ok=true、deterministic=true。评分器reward同时报告replay_success=0、task_failure=1。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频0.00–1.20 s中，MyoLegs以脚接触木质场地地面承重，双腿迈动时身体向前移动；随后身体倾倒。视频全程未见模型接触黑色柱状对手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 1.6,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约1.60 s中，正在前移的MyoLegs身体明显倾斜、躯干与腿的投影逐渐横倒；视频1.60–2.44 s中未恢复直立，仍远离右前方对手。推断：追及过程中发生未修复的失衡，断点为“追及对手”。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 1.6,
+    "log_step": null,
+    "evidence_zh": "观察：视频1.28–2.44 s连续帧中，MyoLegs从直立外形逐渐侧倾、身体下沉，末段明显失去直立支撑姿态，始终没有恢复，也未到达对手。log step 42对交付文件的两次重放均以原生failure结束。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：agent的肌肉控制器未能维持前移中的支撑和平衡。log step 42的交付脚本先跟踪修改后的初始关节姿态，再切换步态参考，并将生成的61步动作保存；同一文件两次重放均失败且末态一致，视频约1.60–2.44 s直接显示倾倒。log step 36、41的会话测试在多组参考相位和增益下也出现骨盆下沉并失败，支持控制能力不足这一归因，但不足以证明某个参数或接触几何是唯一原因。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志，俯视画面中的模型较小，无法精确定位骨盆跨过0.5 m高度阈值的瞬间，或区分失衡究竟由参考轨迹、肌肉力分配还是反馈调节引起，故根因仅标“控制”。评分器reward未报告distance数值，native_metrics中留null；会话测试的best_d不充当评分器指标。log step 51自述的仿真约0.61 s与视频2.48 s属于不同时间尺度，本文所有视频锚点均取自所检视帧。"
+ },
+ "summary_zh": "评分轨迹只前移了一小段便倾倒，始终未追及固定对手，最终完成0/1节点。视频约1.60–2.44 s显示未恢复的失衡，交付文件两次确定性重放均失败，根因支持标为控制不足。缺少逐帧状态日志和评分器距离值，无法进一步确定失衡的具体控制机制。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点（追及对手·倾倒）、平衡现象与控制根因均与视频一致；视频时间均在片内，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:59:04Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--ddgtnpd/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--ddgtnpd/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "986cae30048c93566d2b4d9b89813b56c15fb66051008fd5b3954948d9d0ba33",
+   "video_sha256": "5c8ab98b7c02335f67c3e96aa1368a17ecc4b42930d155e1a5b19e457527d1df",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--stwgf9z.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--stwgf9z.json
new file mode 100644
index 0000000000000000000000000000000000000000..92db9ae329a0f9afa5a6b0b0a55859e89128f615
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--stwgf9z.json
@@ -0,0 +1,164 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-privileged--stwgf9z",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 450,
+   "sim_steps": 450,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 36,
+   "agent_resets": 3775,
+   "agent_sim_steps": 4354,
+   "agent_records": 8,
+   "agent_replays": 3,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.334,
+   "peak_processes": 17,
+   "peak_disk_gb": 0.505,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-privileged--stwgf9z/replay.mp4",
+  "duration_s": 18.04,
+  "fps": 25.0,
+  "frames": 451,
+  "sha256": "98bfd85d121a8ab07760ef5f6647b16612ce13213db4d60d2506f563b3b295dc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [
+     [
+      18.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s，白色MyoLegs肌肉骨骼模型位于场地中部，带浅蓝圆顶的黑色柱状对手位于其右前方，两者明显分离；视频 0–17 s 的逐秒画面显示模型迈步并逐渐转向、接近固定对手。"
+     },
+     {
+      "t": 17.96,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 17.00–18.00 s，模型继续迈步靠近对手；视频 17.96 s 对手圆顶仍为浅蓝色，视频 18.00 s 末帧圆顶变绿，模型骨盆已到对手附近，身体未倒伏。成功参照视频 9.64 s 末帧也出现圆顶变绿，仅用作成功提示的外观对照。"
+     },
+     {
+      "t": 18.0,
+      "log_step": 38,
+      "text_zh": "观察：log step 37 中agent脚本生成并保存最终450步轨迹，输出成功、终点水平距离0.49948371705143674米、骨盆高度0.8652807316460549米及仿真 4.50 s；log step 38 中基准重放工具对该交付文件两次重放均报告success=true、steps=450且终态哈希相同。评分器字段success=1、replay_success=1、deterministic=1，与视频 18.00 s 的绿色成功提示一致。"
+     },
+     {
+      "t": 18.0,
+      "log_step": 38,
+      "text_zh": "推断：结合视频 18.00 s 的成功提示、最终交付文件的重放验收和评分器成功字段，追及条件在视频末帧成立，区间记为[18.00 s, 视频结束]；此时刻直接取自视频帧，不将agent用时或日志中的仿真 4.50 s 当作视频时间。评分器reward未单列distance数值，因此native_metrics的distance保留null。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0–18.00 s，MyoLegs双腿交替迈步，足部落在木质场地地面上支撑身体并推动骨盆接近对手；实际支承交互为脚与地面。视频未显示可确认的身体撞击对手，追及验收也不要求实际触碰对手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "MyoLegs通过双脚支撑、迈步和转向接近固定对手，在视频 18.00 s 末帧触发圆顶变绿的成功提示，追及对手节点完成，最终为1/1。log step 38 对最终450步交付轨迹的两次重放均成功，与评分器成功结果及视频相符，无失败断点。未提供逐帧状态日志，无法独立量测精确水平距离；评分器未单列distance数值，故原生距离指标保留null，节点达成时刻取自视频成功提示。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "opponent_tagged 在视频 18.00 s 圆顶由浅蓝变绿，与评分器 success=1 及重放验收一致；接触标签（腿/躯干·支撑）合理，无失败断点。无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:01:50Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--stwgf9z/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--stwgf9z/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b62fa901dea06474eeda780ebb89ffa68354feac2ccd32522bd153fa2058eefa",
+   "video_sha256": "98bfd85d121a8ab07760ef5f6647b16612ce13213db4d60d2506f563b3b295dc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--zpujabs.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--zpujabs.json
new file mode 100644
index 0000000000000000000000000000000000000000..7d417994dd7fe179649b67420a4c0796660b0c5b
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-privileged--zpujabs.json
@@ -0,0 +1,184 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-privileged--zpujabs",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 88,
+   "sim_steps": 88,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 331,
+   "agent_resets": 6213,
+   "agent_sim_steps": 25214,
+   "agent_records": 1,
+   "agent_replays": 18,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 4.942,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.269,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-privileged--zpujabs/replay.mp4",
+  "duration_s": 3.56,
+  "fps": 25.0,
+  "frames": 89,
+  "sha256": "0fd5f5a1be09961389a59a0c827a05e53a61fb2be5734698daf59620c69fb555"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.0–3.52 s的俯视画面中，白色MyoLegs模型从场地中部向前移动，带浅蓝圆顶的黑色柱状对手始终在其右前方，两者地面投影一直明显分离；视频3.52 s结束时模型已后仰下沉，仍未靠近对手。"
+     },
+     {
+      "t": 3.52,
+      "log_step": 438,
+      "text_zh": "观察：log step 438的诊断脚本从初态逐行重放最终88步交付文件，读取pelvis世界位置和对手位置；控制步索引87处骨盆高度0.486 m，水平距离3.620 m。log step 441随后对同一交付文件两次重放均success=false、steps=88、末态哈希相同，与视频末尾下沉且远离对手一致。"
+     },
+     {
+      "t": 3.52,
+      "log_step": null,
+      "text_zh": "推断：本次全程未进入0.5 m追及阈值。作为视图参照，成功oracle视频9.64 s中白色模型骨盆已紧邻黑色对手，圆顶变绿；本次视频3.52 s无此接近关系，且评分记录success=0、task_failure=1。oracle仅用于理解成功外观，不作为本次进度证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频0.0–2.8 s，MyoLegs双脚交替落在木质场地上支承身体并迈步移动；随后身体后仰下沉。全程未见模型与柱状对手实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 2.8,
+   "log_step": 438,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约2.8–3.52 s，正在迈步接近对手的MyoLegs模型逐渐后仰、骨盆下沉，未恢复直立，结束时与对手仍明显相隔。log step 438重放最终交付文件的末控制步索引87处骨盆高度为0.486 m；log step 441两次确定性重放均在88步失败。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 2.8,
+    "log_step": 438,
+    "evidence_zh": "观察：视频2.8–3.52 s逐帧画面中，MyoLegs身体逐渐后仰并降低，末帧不再保持正常直立支撑姿态，追及尚未完成；log step 438对最终交付轨迹的重放读取末控制步骨盆高度0.486 m，低于DAG所述平地跌倒阈值0.5 m。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 439,
+    "evidence_zh": "推断：log step 439所示交付生成脚本用关节位置/速度反馈、躯干俯仰与侧倾力矩及有界肌肉分配驱动行走，log step 440执行该脚本保存88步，log step 441两次重放一致失败；结合视频2.8–3.52 s的后仰下沉及log step 438的末步骨盆高度0.486 m，表明该肌肉行走控制未能维持支撑和平衡。证据不足以区分目标步态、力矩分配或反馈增益中的具体缺陷。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志，俯视低分辨率画面无法精确恢复足底接触力或失衡起始时刻；log step 438提供的是最终交付文件的稀疏重放诊断，可佐证末态，但不足以定位具体控制子类。评分器reward未报告distance数值，因此native_metrics.distance留null，诊断脚本读取的3.620 m仅作辅助证据。",
+  "unmet": {
+   "opponent_tagged": false
+  }
+ },
+ "summary_zh": "MyoLegs模型尝试迈步接近固定对手，但视频0.0–3.52 s始终未达到追及距离，唯一节点“追及对手”最终未完成（0/1）。视频约2.8 s起后仰下沉，最终交付文件的末步诊断骨盆高度为0.486 m，两次确定性重放均失败，支持平衡丧失及控制层归因。缺少逐帧接触力与控制误差记录，无法确定更具体的控制缺陷。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点 opponent_tagged 未达成、final=false、breakpoint 与平衡/控制归因均与视频、log step 438/441 一致（末步骨盆高度 0.486 m<0.5 m、距离 3.62 m，两次确定性重放失败）；未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:06:46Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--zpujabs/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-privileged--zpujabs/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b562f9642912a3990e09b6f3fac039676f0b7fb5ac5eafc4b6ceaf4f933156b3",
+   "video_sha256": "0fd5f5a1be09961389a59a0c827a05e53a61fb2be5734698daf59620c69fb555",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--4n8f8yi.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--4n8f8yi.json
new file mode 100644
index 0000000000000000000000000000000000000000..7dfe21e9df6552c5bbb1b155d3996e8e22d64164
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--4n8f8yi.json
@@ -0,0 +1,184 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-standard--4n8f8yi",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "physics_steps": 0,
+   "steps": 0,
+   "calls": 2,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 0,
+   "elapsed_s": 2454.9,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.467,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.371,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-standard--4n8f8yi/replay.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "a894b7592370159d3a1ba5b7e128c6a55a1080d916a5214612192d2dc7ebf615"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：评分视频仅含视频 0.00 s 的一帧（总长0.04秒）；白色MyoLegs模型位于场地中央附近，带浅蓝圆顶的黑色柱状对手在其右下方，两者地面投影明显分离，未达到追及距离。"
+     },
+     {
+      "t": 0.0,
+      "log_step": 64,
+      "text_zh": "观察：评分器记录 n_actions=0、sim_steps=0、episodes=1、resets=0、live_success=0；agent 在 log step 64 明确自述未在服务中运行控制器。推断：评分回合始终停在初态，追及步骤没有启动；日志中的本地测试不构成该回合进度。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 0.0,
+   "log_step": 64,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0.00 s 的唯一画面中，MyoLegs与对手仍远离；评分器记录零动作、零仿真步。agent 在 log step 64 自述因本地控制器测试持续跌倒，决定不在服务中启动控制器。推断：评分回合的追及步骤从未开始，断点位于首步前。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 64,
+    "evidence_zh": "观察：视频 0.00 s 为初始分离画面，完整评分视频仅这一帧；评分器 n_actions=0、sim_steps=0、resets=0。agent 在 log step 64 自述没有向现场服务运行控制器。推断：MyoLegs未开始追及对手；本地测试的跌倒不是评分回合中的现象。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 60,
+    "source_zh": "日志",
+    "evidence_zh": "观察：agent 的本地仿真测试在 log step 19、30、60 分别于本地仿真 0.61 s、1.32 s、0.47 s 输出 fallen=True、win=False；log step 64 中 agent 自述控制器无法稳定落地，因此没有在服务中启动。推断：控制器尚未形成可用的稳定行走控制，促使agent放弃现场执行；依据是会话测试及其与停止的明确关联，不是从零进度反推。"
+   }
+  ],
+  "gap_zh": "观察：inputs/包含评分视频和完整会话日志，但没有逐帧状态日志；评分视频只有初始一帧，评分器未报告原生distance数值。推断：本地测试支持控制能力不足的宽泛归因，但不能验证该控制器在评分回合中的落地表现，也不能确定肌肉分配、反馈或具体参数中的哪一项是最终原因，故不细分控制子类。"
+ },
+ "summary_zh": "评分回合停留在初态，追及对手未启动，最终完成0/1节点；唯一视频帧中两者明显分离，评分器记录零动作、零仿真步。断点为追及对手的首步前，现象为没尝试；本地测试持续跌倒及agent最终说明支持控制能力不足促使其未启动现场执行的归因。没有逐帧状态日志或现场运动，具体控制机制无法核实，原生距离指标也未报告。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "评分回合0动作、0仿真步，视频仅一帧初态；节点未达、断点首步前/没尝试、根因控制（来自会话本地测试，按FAIL-22）均有证据支持，时钟与推断标注无误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:15Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--4n8f8yi/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--4n8f8yi/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9cee0f3839504a5e79506e1f3682971a8ffd01afd5701f5ff7b7353400df55b7",
+   "video_sha256": "a894b7592370159d3a1ba5b7e128c6a55a1080d916a5214612192d2dc7ebf615",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--c6usgpu.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--c6usgpu.json
new file mode 100644
index 0000000000000000000000000000000000000000..448e05c2f4a4c064fa13419beff0a411a58280b7
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--c6usgpu.json
@@ -0,0 +1,195 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-standard--c6usgpu",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 99,
+   "sim_steps": 99,
+   "physics_steps": 990,
+   "steps": 294,
+   "calls": 15,
+   "episodes": 5,
+   "resets": 4,
+   "reset_budget": 10,
+   "sim_steps_total": 294,
+   "elapsed_s": 2863.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.625,
+   "peak_processes": 15,
+   "peak_disk_gb": 0.582,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-standard--c6usgpu/replay.mp4",
+  "duration_s": 4.0,
+  "fps": 25.0,
+  "frames": 100,
+  "sha256": "7748ba0573867e662a34db9f1fd5867948019be2058c9b4efa0f48431ead38b2"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s，白色MyoLegs位于场地中部，浅蓝圆顶黑柱对手在其右前方，两者地面投影明显分离，追及条件起初不成立。"
+     },
+     {
+      "t": 2.0,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–3.96 s，模型双腿迈动并向画面下方移动，但骨盆始终与右下方对手有明显间隔；视频 2.00 s 尚未走到对手旁，视频 3.96 s 身体已大幅倾倒，仍未接近到成功参考末段所示的贴近位置。"
+     },
+     {
+      "t": 3.96,
+      "log_step": 152,
+      "text_zh": "观察：log step 151 的最终部署脚本重置后执行肌肉序列，log step 152 的服务状态为 episode=5、control_steps=99、done=true，返回未追及/出界/跌倒的合并失败原因，末次骨盆root_z=0.48 m；评分字段 graded_by_live=1、live_success=0、n_actions=99 与该最后回合相符。视频 3.96 s 模型仍远离对手，故本节点从未达成；此前本地测试的运动不计入进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0.40–2.00 s，MyoLegs双腿迈动，脚在木质地面落下并支承身体；随后视频约 2.80–3.96 s 身体倾倒、支承失稳。未见模型接触对手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 2.8,
+   "log_step": 152,
+   "where": "操作中",
+   "evidence_zh": "观察：模型在追及移动中，于视频约 2.80 s 已出现明显倾倒，视频 3.00–3.96 s 倾斜继续扩大、身体下沉，始终未靠近右下方对手。log step 152 的最后现场回合返回失败，末次骨盆root_z=0.48 m，之后未恢复或重新执行；断点定位为正在尝试的追及对手节点。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 2.8,
+    "log_step": 152,
+    "evidence_zh": "观察：视频约 2.80–3.96 s，MyoLegs躯干由倾斜转为近水平，骨盆下沉，未能保持站立支承；log step 152 记录最后回合末次root_z=0.48 m，低于任务文本规定的跌倒阈值0.5 m，并返回done=true及失败原因。该倾倒发生在尚未追及对手时。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 145 的seq函数仅按时间和固定参数分配各肌肉激活；log step 151 的最终脚本在执行前生成全部指令，执行中读取的本体状态仅用于打印和判断回合结束，没有依据骨盆高度或身体倾斜调整后续指令。结合视频约 2.80–3.96 s 持续倾倒及log step 152 的低骨盆高度，所交付的肌肉控制序列未能维持行走支承，导致追及中断；证据不足以再区分具体参数或控制子类。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志或足底接触力记录，低分辨率俯视视频不能给出精确骨盆距离、接触部位及倾倒起点；断点时间取视频约 2.80 s 的可见明显倾倒处。评分字段未报告distance，因此原生距离指标置null；log step 149 的3.17 m来自本地测试，不是本次评分回合的距离。最终服务回合与本地测试为何分歧无法确定，不归因于仿真或环境；服务控制步的仿真时钟不直接当作视频时钟。"
+ },
+ "summary_zh": "评分采用最后的第5回合：MyoLegs实际迈步并以脚支承地面，但全程未追及固定对手，节点最终为0/1。视频约 2.80 s 开始明显倾倒，至视频 3.96 s 身体下沉，log step 152 的末次骨盆高度为0.48 m；最终按预生成肌肉序列执行的控制未能维持支承。无逐帧状态和接触力记录，无法细分控制失稳机制或给出精确末端距离。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点（追及中跌倒，平衡）与根因（开环肌肉序列，停在控制）均与视频帧和log step 152的服务读数一致；未发现需要修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:01:27Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--c6usgpu/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--c6usgpu/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0a2d1b131c25c2903c73e181c0b22ba9519edfdfc79e0a46a32fa04d1ad0c30a",
+   "video_sha256": "7748ba0573867e662a34db9f1fd5867948019be2058c9b4efa0f48431ead38b2",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--kscoqvt.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--kscoqvt.json
new file mode 100644
index 0000000000000000000000000000000000000000..ff9991e02ac516469458a13298b785d1b7136ec2
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--kscoqvt.json
@@ -0,0 +1,179 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-standard--kscoqvt",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "physics_steps": 0,
+   "steps": 0,
+   "calls": 1,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 0,
+   "elapsed_s": 380.2,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.439,
+   "peak_processes": 6,
+   "peak_disk_gb": 0.325,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-standard--kscoqvt/replay.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "a894b7592370159d3a1ba5b7e128c6a55a1080d916a5214612192d2dc7ebf615"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：本次视频只有视频 0.00 s 的初始帧（全片时长0.04秒），白色模型位于场地中央，浅蓝圆顶黑色柱状对手位于右下方，两者地面投影明显相隔，未呈现追及到0.5米以内的关系。"
+     },
+     {
+      "t": null,
+      "log_step": 5,
+      "text_zh": "观察：log step 5 的机器人服务初始观测记录 control_steps=0、physics_steps=0、sim_time_s=0；run.json 的评分字段 n_actions=0、sim_steps=0、resets=0、live_success=0，完整会话未向服务提交动作。"
+     },
+     {
+      "t": null,
+      "log_step": 23,
+      "text_zh": "观察：log step 23 的最后消息为“This request was blocked by Gemini's filters”；此前日志是初始观测、源码检查、对手位置估计和本地模型测试，无已执行并验证的现场追及结果。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "观察：评分字段 n_actions=0、sim_steps=0；会话在 log step 23 被提供方过滤器中断，未交付运动轨迹，录像仅保存初始画面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "unknown",
+   "t": null,
+   "log_step": 23,
+   "where": "首步前",
+   "evidence_zh": "观察：log step 23 明确报告 Gemini 过滤器拦截；评分字段 n_actions=0、sim_steps=0，视频 0.00 s 唯一帧仍是模型与对手分离的初态。推断：现场追及尚未开始且未形成有效结果便遭提供方中断，按外因中断处理，无法定位物理失败步骤。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "观察：inputs/ 中有完整会话记录和仅一帧、时长0.04秒的本次录像，无逐帧状态日志；log step 23 是提供方过滤器拦截信息，未交付任何现场动作，评分记录同时标记 timeout=1。推断：这次中断应重试并排除在 agent 失败分析之外；保持原始 success=0，但不将外因记为没尝试或能力根因。本次没有可供判断步态、平衡或控制质量的运动过程，distance 原生指标也未报告，故现象和根因均留空。"
+ },
+ "summary_zh": "观察：本次视频 0.00 s 唯一帧中模型与对手仍明显分离，评分记录的动作数和仿真步数均为零，追及对手节点未达成（0/1）。观察：log step 23 明确显示 Gemini 过滤器拦截，会话只完成初始观测和本地检查；推断：属于有效现场动作开始前的外因中断，应重试并排除在 agent 失败分析之外。观察：没有运动录像或状态日志，且评分器未报告 distance，无法评价具体运动失败现象或能力根因。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标签与证据一致：视频仅一帧初态，n_actions=0，log step 23 为 Gemini 过滤器拦截，未交付任何动作；节点未达成，现象和根因留空并有 gap_zh，符合外因中断规则。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:02:39Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--kscoqvt/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--kscoqvt/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "7332057f7502705cfdaa3fbd6d0384697da5957460d5b0573daea53d38351b68",
+   "video_sha256": "a894b7592370159d3a1ba5b7e128c6a55a1080d916a5214612192d2dc7ebf615",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--kwb4otb.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--kwb4otb.json
new file mode 100644
index 0000000000000000000000000000000000000000..7c3824e1cd4a533babdd4c58a9d20ad015edf557
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--kwb4otb.json
@@ -0,0 +1,190 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-standard--kwb4otb",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 74,
+   "sim_steps": 74,
+   "physics_steps": 740,
+   "steps": 529,
+   "calls": 39,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 579,
+   "elapsed_s": 422.4,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.648,
+   "peak_processes": 9,
+   "peak_disk_gb": 1.167,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-standard--kwb4otb/replay.mp4",
+  "duration_s": 3.0,
+  "fps": 25.0,
+  "frames": 75,
+  "sha256": "8f441038d66b29a055eea09799b197aba50c415bdee550927a6d1a02f4183ddb"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 白色 MyoLegs 位于场地中央附近，带浅蓝圆顶的黑色柱状对手位于画面右下方，两者明显分离。已检查视频 0.0–2.96 s 的全部 75 帧，模型始终远离对手，未出现骨盆投影靠近对手至阈值内的画面。"
+     },
+     {
+      "t": 2.96,
+      "log_step": 61,
+      "text_zh": "观察：视频 2.0–2.96 s 模型的躯干翻转、双腿蜷曲，仍在对手左上方较远处；log step 61 的服务状态为 episode=11、control_steps=74、done=true，reason 为未追及、出界或跌倒的合并失败说明。评分器 live_success=0、graded_by_live=1，与该末回合未追及的画面一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0.0–1.2 s，MyoLegs 以脚部接触木质地面支承身体，双腿改变姿态并移动；随后模型失衡下沉。整个视频未见模型接触柱状对手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 1.8,
+   "log_step": 60,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 1.8 s，正在以双腿运动的 MyoLegs 已明显倾斜、转体，随后至视频 2.96 s 持续翻转下沉，始终未靠近右下方的对手；log step 60 记录末回合支撑阶段执行后，在仿真 0.74 s 骨盆高度 z=0.497、done=true。视频时间以所看帧为准，不将该仿真时刻直接当作视频时刻。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 1.8,
+    "log_step": 60,
+    "evidence_zh": "观察：视频 1.8–2.96 s，MyoLegs 躯干从倾斜进一步翻转，身体下沉、双腿蜷曲，没有恢复站立；log step 60 的末回合输出在仿真 0.70 s 记录骨盆高度 0.542，在仿真 0.74 s 记录 0.497 且 done=true，支持失衡跌倒导致回合结束。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：末回合肌肉控制未能建立稳定的腿部支撑。log step 58 说明 agent 选择左右支撑阶段并延长右侧支撑；log steps 59–60 执行该脚本后，输出依次为左支撑、右支撑及保持阶段，骨盆高度从仿真 0.35 s 的 0.828 降至仿真 0.74 s 的 0.497；视频 1.8–2.96 s 同时可见持续翻转下沉。证据支持控制失稳，但不足以区分接触布局、反馈律或具体参数问题。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志；评分器 reward 未报告 distance，故该原生指标为 null。录像只有低分辨率俯视图，无法精确测量骨盆距离或各脚受力；日志中的脚本编辑内容未完整展开，不能把失稳进一步归因于某个肌肉参数或反馈索引。末回合仿真终点为 0.74 s，视频时长为 3.0 s，未建立精确的仿真到视频映射，断点仅按画面作粗略定位。",
+  "unmet": {
+   "opponent_tagged": false
+  }
+ },
+ "summary_zh": "评分采用最后一次重置后的第 11 回合；视频全程未追及固定对手，节点最终为 0/1。MyoLegs 在视频约 1.8 s 已明显失衡，之后翻转下沉，末回合日志也记录骨盆高度降至跌倒阈值以下，根因支持标为控制失稳。缺少逐帧状态及完整末回合控制脚本，不能进一步确认具体控制参数，评分器也未报告最终距离。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点 opponent_tagged=false、断点（opponent_tagged，视频约1.8 s 倾斜失衡）、现象（全身·平衡）与根因（控制，已声明不足以细分）均与视频和 log step 60-61 一致；未发现需修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:00:08Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--kwb4otb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--kwb4otb/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "25d4d5cb15f58504299f270cccda8d2c8cd9e22d30015a20ce09a72fca94fc34",
+   "video_sha256": "8f441038d66b29a055eea09799b197aba50c415bdee550927a6d1a02f4183ddb",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--rn5emks.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--rn5emks.json
new file mode 100644
index 0000000000000000000000000000000000000000..487d9f4282209aaec1c552f7a9668be033e3883e
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--rn5emks.json
@@ -0,0 +1,192 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-standard--rn5emks",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 63,
+   "sim_steps": 63,
+   "physics_steps": 630,
+   "steps": 184,
+   "calls": 143,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 184,
+   "elapsed_s": 2546.4,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.701,
+   "peak_processes": 14,
+   "peak_disk_gb": 1.324,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-standard--rn5emks/replay.mp4",
+  "duration_s": 2.56,
+  "fps": 25.0,
+  "frames": 64,
+  "sha256": "f96dc9a84d9b12267ad5879cdc8c3e29efd3cc6aa90e715976f678f1b82dcbc8"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00 s，白色MyoLegs模型位于场地中央，带浅蓝圆顶的黑色柱状对手位于画面右下方，两者明显分离。逐帧检查视频0.00–2.52 s，模型仅向画面下方移动少许，并逐渐屈腿下沉、倾斜，骨盆始终没有接近对手至0.5米的范围。"
+     },
+     {
+      "t": 2.52,
+      "log_step": 174,
+      "text_zh": "观察：视频2.52 s，模型仍在场内、远离对手；成功参考inputs/oracle/replay.mp4的视频约9.60 s则可见模型紧邻对手。log step 173重置后运行控制脚本，log step 174的服务状态为episode=3、control_steps=63、done=true，reason为未追及、出界或跌倒的合并失败信息；评分字段live_success=0、graded_by_live=1、n_actions=63，与最后回合一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频中MyoLegs以双脚落脚支承木质地面，双腿屈伸尝试推进身体；随后腿部弯曲、身体下沉倾斜。未见模型与柱状对手实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 1.6,
+   "log_step": 174,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约1.20–2.52 s，MyoLegs在推进过程中屈腿下沉，身体逐渐倾斜，未能走向右下方对手附近；视频约1.60 s作为失稳过程的导航锚点。log step 173启动最后的第3回合，log step 174报告该回合在63个控制步后失败结束，之后没有再重置或修复。推断：失败发生在追及对手的行走操作中。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 1.6,
+    "log_step": 174,
+    "evidence_zh": "推断：视频1.20–2.52 s可见MyoLegs腿部屈曲、身体下沉倾斜，且仍在场内并远离对手；log step 174服务报告第3回合失败，任务控制率100 Hz（log step 15），63个控制步只对应仿真0.63 s，尚未达到仿真20 s时限。结合任务的失败条件，符合骨盆过低的跌倒失败，而非正常步态继续行进；视频未延伸到身体完全倒伏。此处仿真时长不作为视频时间。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 1.6,
+    "log_step": 130,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 130的最终脚本读取本体反馈，以骨盆位置与姿态PD、腿部关节PD和交替迈步目标计算肌肉指令，骨盆高度目标至少为0.95米；log step 173实际运行该脚本，log step 174显示第3回合完成63个控制步后失败，视频1.20–2.52 s仍出现下沉倾斜。控制器没有实现其维持支撑并推进的目标，支持控制层面的归因；没有足够的现场跟踪误差或接触力记录来确认具体子类。"
+   }
+  ],
+  "gap_zh": "无逐帧状态日志。现有320×240俯视视频可判读两者始终分离及模型下沉倾斜，但不能直接量出骨盆高度、逐足接触力或精确水平距离；评分器未报告distance，原生失败原因也是多个条件的合并信息。跌倒判读由画面、场内位置、回合控制步数和失败条件共同推断，断点时间仅为视频约1.60 s的粗略锚点。日志中的本地副本测试属于会话诊断，不能当作第3回合的状态轨迹；具体失稳机制未确定。"
+ },
+ "summary_zh": "最后评分的第3回合中，MyoLegs尝试以双脚支撑推进，但始终没有接近固定对手至成功距离，追及对手节点终态为0/1。视频约1.20–2.52 s显示模型屈腿下沉、身体倾斜，结合log step 174的63个控制步后失败状态，判断其在追及过程中失去平衡，最终控制器未能维持有效支撑。没有逐帧状态日志或评分器距离数值，精确骨盆高度及具体控制失稳机制无法确认。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点 opponent_tagged 未达成、final=0 与视频（0–2.52 s 两者始终分离）及服务读数（第3回合 63 控制步 done=true 失败，0.63 s 仿真内不可能超时/出界，故为跌倒）一致；断点、平衡现象、控制根因均有证据且已标为推断，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:08:02Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--rn5emks/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--rn5emks/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "696d86e375e0faf7fff057c566b746c9d701e8aaaca0fd8aaa43ec24f0862481",
+   "video_sha256": "f96dc9a84d9b12267ad5879cdc8c3e29efd3cc6aa90e715976f678f1b82dcbc8",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--tmcuy4h.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--tmcuy4h.json
new file mode 100644
index 0000000000000000000000000000000000000000..9b8e2bf0dd339f2362f32bd69e1ac986c87629af
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--tmcuy4h.json
@@ -0,0 +1,187 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-standard--tmcuy4h",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 86,
+   "sim_steps": 86,
+   "physics_steps": 860,
+   "steps": 922,
+   "calls": 931,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 922,
+   "elapsed_s": 3308.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-standard--tmcuy4h/replay.mp4",
+  "duration_s": 3.48,
+  "fps": 25.0,
+  "frames": 87,
+  "sha256": "b5967619cd3513b8b4d7b977a92b19aea24de82bab8b14eccf7cf3b909724176"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 118,
+      "text_zh": "观察：log step 118 执行最后一次 reset 后开始控制；log step 137、139 的服务状态均为 episode=11、resets_left=0，评分器记录 n_actions=86、live_matches_replay=1。本次视频从白色模型站在场地中部开始，只呈现这一最后回合，不包含前十回合的测试。"
+     },
+     {
+      "t": 2.4,
+      "log_step": 126,
+      "text_zh": "观察：视频 0.00–2.40 s 的俯视画面中，白色模型双腿交替移动，身体略向画面下方推进；黑色柱状、浅蓝圆顶对手一直在其右下方，两者间仍有明显空地。"
+     },
+     {
+      "t": 3.44,
+      "log_step": 139,
+      "text_zh": "观察：已检查视频全程每约 0.20 s 的画面及视频 2.40–3.44 s 的每帧；白色模型末段倾倒，直到视频 3.44 s 仍与对手明显分离，圆顶未变绿。成功参考视频 9.64 s 中模型已经靠到对手旁，圆顶变绿；该参考仅用于辨认达标外观。log step 139 的最后回合服务状态为 done=True、reason=failure，评分器 success=0。"
+     },
+     {
+      "t": 3.44,
+      "log_step": 139,
+      "text_zh": "推断：本视频 87 帧对应初态及最后回合的 86 个控制步，以 25 fps 播放；控制步 k 对应视频 k/25 s，而脚本每步 dt=0.01 对应仿真 k/100 s。log step 139 的 control_steps=86、骨盆高度 0.493 m 与视频 3.44 s 的下沉倾倒末帧吻合，据此核对映射；不能把视频时长当作仿真时长。结合全程可见间距及原生失败结果，“追及对手”从未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0.00–2.40 s，MyoLegs 的双脚在木质地面交替落脚并支承身体，双腿带动本体向前移动；视频 2.80–3.44 s 身体倾倒、支撑未能维持。未见模型与柱状对手实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 2.8,
+   "log_step": 129,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 2.80 s 起白色模型躯干倾斜加重，随后骨盆持续下沉，视频 3.44 s 结束时仍未靠近对手。log step 129、131、135、137、139 的最后回合读数依次给出骨盆高度 0.774、0.711、0.613、0.510、0.493 m，最终服务判负。推断：机器人正在执行追及导航时失衡，且直到回合结束未恢复或完成该步；断点取“追及对手”。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 2.8,
+    "log_step": 129,
+    "evidence_zh": "观察：视频 2.80–3.44 s 的连续画面中，白色模型由较直立姿态变为明显倾斜、压低的姿态，未恢复站立；log step 129 的骨盆高度为 0.774 m，log step 139 为 0.493 m，后者同时报告最后回合 done=True、failure。推断：模型失去平衡，骨盆下降越过任务的 0.5 m 跌倒阈值而终止，未追及对手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：肌肉步行控制未能在支撑切换中稳定本体。log step 123、128、130 显示 agent 在最后回合连续修改摆腿伸膝和接触阶段逻辑；log step 136 又在骨盆过低时添加伸膝与髋力矩恢复指令，但 log step 135、137、139 的实际本体读数显示骨盆仍由 0.613 m 降至 0.510 m、0.493 m，视频 3.20–3.44 s 亦显示倾倒继续。上述证据支持控制未能稳定身体，尚不足以把失败唯一归于接触估计、肌肉模型或某一增益参数。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志；输入仅含远距离俯视视频，足底接触细节及真实接触力无法分辨，因而不能验证控制器接触估计的误差或进一步定位失稳机制。评分器未报告原生 distance 数值，故该指标为 null；未用图像估计替代米制距离。"
+ },
+ "summary_zh": "最后第 11 回合中，MyoLegs 通过双脚支撑尝试向前移动，但一直未靠近固定对手，追及节点始终未达成，最终为 0/1。视频约 2.80–3.44 s 模型失衡下沉，最后回合 log step 139 的骨盆高度为 0.493 m 且服务判负，支持肌肉步行控制未能维持平衡。缺少逐帧状态、接触力和原生距离数值，无法进一步确定失稳的具体控制环节。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标注与视频、日志一致：追及节点未达成（0/1），控制步70对应视频2.8 s（log step 129 的 step 70、骨盆0.774 m）映射成立，断点、平衡现象和控制根因均有证据支持，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:01:29Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--tmcuy4h/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--tmcuy4h/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "011237f53cf6207fff0df0b29957d04883e35304af6b7440c07506eda750d428",
+   "video_sha256": "b5967619cd3513b8b4d7b977a92b19aea24de82bab8b14eccf7cf3b909724176",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--uxe89cz.json b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--uxe89cz.json
new file mode 100644
index 0000000000000000000000000000000000000000..1ed72c6272d04e166e1396105c6eaf0b3a1a7ab3
--- /dev/null
+++ b/phase2/myosuite/chase_tag/trajectories/myosuite-chase-tag-i00-standard--uxe89cz.json
@@ -0,0 +1,195 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-chase-tag-i00-standard--uxe89cz",
+ "taskKey": "myosuite/chase_tag",
+ "dag_sha256": "44d1913cbf93a1cace28fb27d008741b9ec7d5862825d6d9128583801a384d16",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 203,
+   "sim_steps": 203,
+   "physics_steps": 2030,
+   "steps": 686,
+   "calls": 23,
+   "episodes": 4,
+   "resets": 3,
+   "reset_budget": 10,
+   "sim_steps_total": 686,
+   "elapsed_s": 2649.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.711,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.533,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-chase-tag-i00-standard--uxe89cz/replay.mp4",
+  "duration_s": 8.16,
+  "fps": 25.0,
+  "frames": 204,
+  "sha256": "550a6409faa61f4cd3fa8b206008e1b4088b97040cffcec9024a75eb627efc0f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "opponent_tagged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 白色MyoLegs模型位于场地中央，浅蓝圆顶的黑色柱状对手在画面右下方，两者明显分离；视频 2.0、4.0、6.0 s 模型虽有向前位移，骨盆仍未接近对手。"
+     },
+     {
+      "t": 8.12,
+      "log_step": 260,
+      "text_zh": "观察：视频 7.2、7.4、7.6、7.8、8.0、8.04、8.08、8.12 s 模型逐渐倾倒，直到最后一帧仍与对手隔着明显地面间隙；对手圆顶未出现参考成功录像末帧的绿色提示。log step 260 明确重置为 Episode 4，服务最终 control_steps=203、done=True，返回失败原因，服务骨盆读数为(0.481,-1.651,0.494) m；评分器字段 live_success=0、n_actions=203 与该末局相符。"
+     },
+     {
+      "t": 8.12,
+      "log_step": 260,
+      "text_zh": "推断：评分回合是 log step 260 的 Episode 4，前几局测试不计入本轨迹。该局203个控制步加初始画面与视频204帧相符，末帧203/25=视频 8.12 s 的下沉姿态与服务骨盆高度0.494 m相互核对；控制步dt=0.01对应仿真 2.03 s，不能把该仿真时长直接当视频时长。脚本以服务骨盆读数和估计对手坐标计算末端距离3.324 m，结合全程画面始终明显分离，追及距离未达0.5 m。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "推断：视频 2.0、4.0、6.0 s 的落脚与身体位移显示，MyoLegs以双腿和双脚接触木质地面、承重推进骨盆；视频约 7.6 s 起明显倾倒，未见模型接触黑色柱状对手。没有手部或轮式底盘参与。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "opponent_tagged",
+   "t": 7.6,
+   "log_step": 260,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 6.8–8.12 s，正在向对手移动的MyoLegs躯干持续倾斜、骨盆下沉，视频约 7.6 s 已明显倾倒，末帧仍与对手明显分离；log step 260 的 Episode 4 最后服务骨盆高度为0.494 m并返回done=True及失败原因。该局未恢复支撑或完成追及，断点为追及对手。"
+  },
+  "unmet": {
+   "opponent_tagged": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 7.6,
+    "log_step": 260,
+    "evidence_zh": "观察：视频 7.0、7.2、7.4、7.6、7.8、8.0及8.12 s，MyoLegs从倾斜站姿逐渐转为躯干接近横向、骨盆下降的倾倒姿态，没有恢复直立；log step 260 的最终服务骨盆高度0.494 m低于任务0.5 m跌倒阈值，此时回合结束。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 254 可见agent脚本先在本地仿真生成肌肉动作序列，再按最多50步一批提交；服务返回的骨盆读数用于打印距离和检查结束，未用于重算后续动作。log step 259 将生成长度改为205步，log step 260 运行该脚本时，最后一局服务骨盆高度从150控制步的0.874 m降至200控制步的0.560 m，再到203控制步的0.494 m；视频 6.0–8.12 s 同步显示逐渐倾倒，说明所执行肌肉控制未能维持追及过程所需支撑。现有证据只支持控制层面的归因，不能确定具体肌群、反馈增益或接触布局的问题。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志或评分器原生distance数值；log step 260 的3.324 m由agent脚本结合服务骨盆读数与估计对手坐标计算，故native_metrics.distance留空。俯视录像分辨率较低，不能解析逐足接触力或具体关节失稳过程；log step 254 的生成函数中段在输入记录中已被截断，根因不细分为反馈控制、轨迹或逆解或参数调试。"
+ },
+ "summary_zh": "评分采用最后一局 Episode 4（log step 260），MyoLegs以双腿落脚推进，但直到视频 8.12 s 结束仍与固定对手明显分离，追及对手节点全程未达成，最终0/1。视频约 7.6 s 出现明显倾倒，最终服务骨盆高度0.494 m，所执行肌肉控制未能维持支撑。没有逐帧状态日志及评分器原生距离值，具体控制失稳机制无法进一步定位。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "评分回合为 Episode 4（log step 260，203 控制步对应视频 204 帧），追及节点全程未达成，终局骨盆 0.494 m 触发跌倒；断点、平衡现象与停在“控制”的根因均有证据支持，未发现需修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:04:10Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--uxe89cz/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-chase-tag-i00-standard--uxe89cz/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "98d8f191437a047df32ed7a389d6f4d28d8286e1ed352f9470639c5275e034ee",
+   "video_sha256": "550a6409faa61f4cd3fa8b206008e1b4088b97040cffcec9024a75eb627efc0f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--5ybpkdi.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--5ybpkdi.json
new file mode 100644
index 0000000000000000000000000000000000000000..46744b4220826f4c48b21fe3fdc90a9c16ce6704
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--5ybpkdi.json
@@ -0,0 +1,202 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-privileged--5ybpkdi",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 36,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 136,
+   "agent_resets": 21398,
+   "agent_sim_steps": 472278,
+   "agent_records": 0,
+   "agent_replays": 3,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.623,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.841,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-privileged--5ybpkdi/replay.mp4",
+  "duration_s": 1.48,
+  "fps": 25.0,
+  "frames": 37,
+  "sha256": "7982fccadcf4e9b1ee5a5805ffde0d4616a1a433349eef123530f7fcec0249f9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.4,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–1.44 s 的全部37帧；MyoHand手指屈曲包围飞机的机身及翼根，飞机只在桌面附近小幅移动和改变朝向，没有出现随手在空中飞行的完整过程。"
+     },
+     {
+      "t": 1.44,
+      "log_step": 223,
+      "text_zh": "观察：视频 1.04–1.44 s 飞机仍贴近桌面，手附近的绿色标记逐渐远离飞机；log step 222 记录提交文件含99行动作，log step 223 对该文件两次重放均仅执行36个控制步，原生终止原因均为物体距参考或手腕超过0.25米，且终态哈希一致；评分器字段sim_steps=36、replay_success=0与提前结束一致。"
+     },
+     {
+      "t": null,
+      "log_step": 223,
+      "text_zh": "推断：该单一event节点要求走完99个控制步，提交文件实际只重放36步便终止，结合视频中没有完成空中运动，飞机循迹飞行未达成；没有把会话中其他候选轨迹的局部成绩计入本次进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 0.32–1.44 s，MyoHand单手弯曲的手指从机身及翼根周围包围、夹持玩具飞机，使其小幅移动；飞机未被带入完整的空中飞行。此处记录可见合手接触，不据此声称稳定承载或成功抬起。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 1.08,
+   "log_step": 223,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 1.08 s 起，MyoHand仍在飞机周围弯曲手指，飞机停留在桌面附近，未随远离的绿色标记升空；视频直到 1.44 s 仍未恢复。log step 223 的提交文件重放在第36控制步因原生距离条件终止，未走完99步。断点锚定可见偏离开始明显的画面，不将控制步或agent工作时间换算成视频时间。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.08,
+    "log_step": 223,
+    "evidence_zh": "观察：视频 1.04–1.44 s，机器人手指继续围住飞机，但飞机只在桌面附近移动、转动，没有跟随所需的空中飞行过程；绿色标记与飞机明显分离。log step 223 对最终文件的两次重放均因物体距参考或手腕超过0.25米提前终止。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 194–195 的会话搜索用前臂目标偏移与肌肉命令优化持续对向接触，输出虽报告接触评分提高，飞机相对初态的竖直位移仍接近零；这些仅作控制问题的会话证据。视频 0.40–1.44 s 中，交付动作也只能使手指围住飞机并小幅移动，未产生所需的升空运动；结合log step 223 的一致重放失败，支持控制层失败，尚不能确定具体控制子类。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功示范；评分器reward未报告track_pos_err、track_rot_err、track_ok_fraction及track_max，故四项原生指标均为null，未借用其他会话候选的数值。视频可见合手接触和未完成飞行，但无法精确确认各指接触力、承载分布及每步误差；不能进一步区分抓取和接触设计、轨迹或逆解、反馈控制。"
+ },
+ "summary_zh": "评分轨迹中，MyoHand合手接触飞机并使其小幅移动，但飞机始终贴近桌面，未完成空中循迹，唯一节点未达成（0/1）。视频约 1.08 s 后偏离明显，最终提交在两次重放中均于第36控制步触发原生距离终止；证据支持控制层失败。没有状态日志和成功示范，四项精确循迹指标未报告，无法细分控制根因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、位姿不准现象与控制根因（停在大类并写明推断）与视频和评分器字段一致，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:55:04Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--5ybpkdi/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--5ybpkdi/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9ba7f87bf4566fa1396f871b029667f1a1671276b278500da226a1232f0971dc",
+   "video_sha256": "7982fccadcf4e9b1ee5a5805ffde0d4616a1a433349eef123530f7fcec0249f9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--biszf8k.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--biszf8k.json
new file mode 100644
index 0000000000000000000000000000000000000000..2283d5909cb2390a1847b68cd33b9391105711e6
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--biszf8k.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-privileged--biszf8k",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 37,
+   "sim_steps": 37,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 348,
+   "agent_resets": 1103,
+   "agent_sim_steps": 10390,
+   "agent_records": 4,
+   "agent_replays": 10,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.583,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.269,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-privileged--biszf8k/replay.mp4",
+  "duration_s": 1.52,
+  "fps": 25.0,
+  "frames": 38,
+  "sha256": "f92ac7f23cd0d309deb34693826910b46dcbb6e09fce1bdb55695e4af799db1c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.96 s 的两个视角中，MyoHand手指贴近并触及银色玩具飞机上侧，手部小幅调整；飞机一直贴着桌面，未见被夹住带起。已检查全部38帧。"
+     },
+     {
+      "t": 1.08,
+      "log_step": null,
+      "text_zh": "观察：视频 1.04–1.48 s，绿色参考标记向上、向外移动，银色飞机仍在桌面原处附近，未随手形成空中运动或跟随参考。"
+     },
+     {
+      "t": 1.48,
+      "log_step": 442,
+      "text_zh": "观察：log step 441生成最终37行、45维动作并重放，track_ok=26、终态哈希a07d006b9acee9f6；log step 442官方评分器两次重放同一提交均在37步失败，原因是物体距参考或手腕超过0.25 m，哈希一致。视频末帧（视频 1.48 s）飞机仍在桌面；未完成DAG要求的99步循迹。"
+     },
+     {
+      "t": null,
+      "log_step": 438,
+      "text_zh": "观察：log step 438对已保存的交付动作逐步重放，零基控制步26位移误差43.14 mm，控制步27为60.72 mm，控制步36为289.88 mm、转动误差0.5189 rad并终止，飞机质心高度始终约17–18 mm。这是交付文件诊断，log step 435和442的重放哈希相同；不把这些控制步或agent工作时间换算成视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–1.48 s，MyoHand单手指端在飞机上侧贴触、擦动；未见手指从相对两侧夹住飞机，也未把飞机带离桌面，因此不记抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_rot_err",
+   "value": null,
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok_fraction",
+   "value": null,
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 1.04,
+   "log_step": 442,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 1.04 s 起，参考标记开始明显离开桌上飞机，而MyoHand手指没有夹持、带起飞机；至视频 1.48 s 均未补成抓取或空中循迹。log step 442的官方重放在37步以物体距参考或手腕超过0.25 m失败。断点为正在尝试的飞机循迹飞行，视频时间仅为未能带起飞机的粗略导航锚点。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 1.04,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–1.48 s，MyoHand指端虽触碰飞机上侧，却没有形成能把飞机带起的包围或对向夹持；视频 1.04–1.48 s 参考向上离开时飞机仍贴桌。缺的是承重抓握接触，未见已持起后再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 440的最终生成脚本每步用rest_action初始化45维动作，只改前6个前臂执行器，手指肌肉保持静息动作；log step 441生成的37行动作与log step 442两次失败重放的终态哈希一致。结合视频 0.00–1.48 s 指端停留于飞机上侧、飞机始终未被带起，所提交控制只调整前臂而未形成承重夹持，接触安排不足以将前臂运动传给飞机。没有成功抓取示范或已验证的替代抓法，替代可行性未知。"
+   }
+  ],
+  "gap_zh": "没有状态日志及成功示范，无法核对逐步接触力或验证替代抓法。run.json的reward及官方重放结果未报告DAG四项原生指标的数值，native_metrics均置null；log step 438含交付文件的舍入诊断误差，只作辅助证据。视频不能精确给出厘米、弧度误差或达标比例；断点时间依据可见的参考离开而粗定，未将20毫秒控制步映射到25 fps视频。"
+ },
+ "summary_zh": "提交轨迹只执行37步，飞机始终留在桌面，飞机循迹飞行节点未完成，最终为0/1。视频约 1.04 s 起参考明显离开，手指未形成承重夹持；最终脚本只控制前臂、手指保持静息动作，支持抓取和接触设计不足，官方两次重放均以距离超限失败。无状态日志或成功示范，精确接触力和替代抓法尚不可确认，评分器未报告的原生指标保留为空。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注与视频、日志一致：飞机始终留在桌面，节点未达成，断点、现象（没建立接触）和根因（抓取和接触设计，推断标注）有证据支持，时钟未换算。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:53:53Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--biszf8k/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--biszf8k/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "7dcdf517919e07f91e7295a7484e2b3a78e5c82b8774ef847f2778d9b2dde086",
+   "video_sha256": "f92ac7f23cd0d309deb34693826910b46dcbb6e09fce1bdb55695e4af799db1c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--efg8rkx.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--efg8rkx.json
new file mode 100644
index 0000000000000000000000000000000000000000..67726643488e2e407a345d808a47aaaf11c9cfb3
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--efg8rkx.json
@@ -0,0 +1,205 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-privileged--efg8rkx",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 36,
+   "sim_steps": 36,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 52,
+   "agent_resets": 319,
+   "agent_sim_steps": 273,
+   "agent_records": 1,
+   "agent_replays": 12,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.979,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.985,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-privileged--efg8rkx/replay.mp4",
+  "duration_s": 1.48,
+  "fps": 25.0,
+  "frames": 37,
+  "sha256": "0e528e76beb57fd29c0c0da83095ac5187dc0186a33ec1bdfa095071f36bc97b"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 玩具飞机在桌面上，MyoHand单手位于飞机上侧；初始接近目标不等于完成整段循迹。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频 0.00–1.44 s 的全部37帧；手指在飞机上侧屈曲、触拨，但飞机一直留在桌面，没有随手形成空中位移和转向；视频 0.96–1.44 s 仍未带起飞机。"
+     },
+     {
+      "t": 1.44,
+      "log_step": 128,
+      "text_zh": "观察：视频 1.44 s 末帧飞机仍在桌面；log step 126 生成并保存36行控制指令，服务返回done=True、track_steps=99、track_ok=24及原生距离终止原因。log step 128 对该交付文件的两次官方重放均在控制步36失败且终态哈希相同，评分器reward也报告n_actions=36、replay_success=0、deterministic=1。未完成99步循迹；不将控制步时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–1.44 s，MyoHand单手指端在玩具飞机机翼及机身上侧触拨，飞机仍由桌面承托；未见拇指与其余手指形成对向夹持，实际接触不记为抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.96,
+   "log_step": 128,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.96 s，MyoHand仍在飞机上侧屈指，飞机留在桌面；至视频 1.44 s 仍无有效带起。log step 126 保存的交付只有36个控制步，log step 128 的官方重放确认该文件因飞机距参考或手腕超过原生阈值而提前终止，飞机循迹飞行始终未完成。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.96,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–1.44 s 中，MyoHand指端虽触拨飞机上侧，但未包住或夹住飞机，飞机始终留在桌面。所需的承载抓持从未建立，未见先提起再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "推断：视频 0.64–1.44 s 的手指作用位置集中在飞机上侧，未形成能带起飞机的对向承载接触；log step 125 的交付生成脚本直接跟踪参考手臂及手指关节角，没有建立抓持的确认或针对飞机接触的修正，log step 126 随即记录提前终止。结合飞机始终未离桌，当前抓持接触布局未奏效；未验证可行替代，不能由会话搜索失败断言任务无解。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功示范，无法确定具体指骨接触对、接触力及可行替代抓法。评分器reward未单独报告DAG四项原生指标，native_metrics值均留null；log step 126 仅给出所保存运行在控制步36终止时的服务读数track_pos_err=0.26702 m、track_rot_err=0.569 rad、track_ok=24，不能据此补出99步完整达标比例或精确历史最大误差。视频逐帧检查足以判断未带起和未完成，不作仿真与视频时钟的等速假设。"
+ },
+ "summary_zh": "观察：评分视频全段中，MyoHand仅在玩具飞机上侧触拨、屈指，飞机始终留在桌面，飞机循迹飞行未完成，节点终态为0/1。推断：当前接触布局未建立有效抓持，无法带动飞机飞行；log step 128 的两次官方重放均在控制步36原生终止，支持物理失败而非交付不一致。缺少状态日志、成功示范及评分器完整循迹指标，无法细分接触力或验证替代抓法。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频 0–1.44 s 飞机始终在桌面（绿球为参考标记），节点未完成、断点、接触（推/扫非抓握）、现象与根因均有证据支持；时钟未把控制步换算成视频时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:50:12Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--efg8rkx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--efg8rkx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ddde06e13533a14be83be13ba4e9ed58f26953ac5e2b9e18102830f024759f05",
+   "video_sha256": "0e528e76beb57fd29c0c0da83095ac5187dc0186a33ec1bdfa095071f36bc97b",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--p6hl228.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--p6hl228.json
new file mode 100644
index 0000000000000000000000000000000000000000..1bbef6e902560a8b2881edeea2fcd9969c77d2e7
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--p6hl228.json
@@ -0,0 +1,215 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-privileged--p6hl228",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 25,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 104,
+   "agent_resets": 247,
+   "agent_sim_steps": 6776,
+   "agent_records": 3,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.998,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.257,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-privileged--p6hl228/replay.mp4",
+  "duration_s": 1.04,
+  "fps": 25.0,
+  "frames": 26,
+  "sha256": "061e8136c533839b105b25408e7ae7016228e86a7929518372e64e95e55bf6ff"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00 s的两个视角中，单只红色肌腱骨骼手位于飞机上方，飞机靠近桌面；初态相近不代表整段参考运动已完成。"
+     },
+     {
+      "t": 0.76,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频0.00–1.00 s（全部26帧）；手指逐渐弯曲，视频0.60–0.80 s拇指与其余弯曲手指从两侧包围飞机机身，飞机被带动横移，但仍贴近桌面。"
+     },
+     {
+      "t": 0.88,
+      "log_step": null,
+      "text_zh": "观察：视频0.80–1.00 s，飞机仍在红色手指下方、靠近桌面，绿色参考标记与飞机的间隔持续拉大；末帧未出现飞机随参考在空中飞行的结果。"
+     },
+     {
+      "t": null,
+      "log_step": 135,
+      "text_zh": "观察：log step 135运行最后一次生成脚本并保存交付文件，输出在控制步25提前终止，原因是飞机距参考或手腕超过0.25 m；该会话报告控制步20位移误差0.0597 m，控制步25位移误差0.25538 m、转动误差0.2737 rad、track_ok=19。评分器reward独立报告sim_steps=25、n_actions=99、task_failure=1、deterministic=1。"
+     },
+     {
+      "t": 1.0,
+      "log_step": 135,
+      "text_zh": "推断：log step 134的脚本将提前结束的动作以最后一行补至99行，log step 135保存后无后续覆盖记录；它的25步终止与评分器及视频短暂的桌面附近运动一致，因此认定为本次交付来源。结合视频中未完成的空中运动与评分器实际只执行25步，完整99步循迹节点从未达到；日志数值是生成会话的佐证，未将agent工作时间或控制步的仿真时间当作视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频0.60–1.00 s，MyoHand单只手的拇指与其余弯曲手指从相对两侧包围、夹住玩具飞机机身并带动其移动；前段手指先从上方接触、压擦飞机表面。此处记录可见的合手接触，不据此认定飞机完成抬升或循迹；半透明白色手形是参考显示，不是第二只操作手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.8,
+   "log_step": 135,
+   "where": "操作中",
+   "evidence_zh": "观察：视频0.80–1.00 s，红色手指合拢并带动飞机在桌面附近移动，飞机与绿色参考标记明显分离，直至视频结束没有修复。log step 135最后保存的生成会话在控制步25因飞机距参考或手腕超过0.25 m而终止，评分器实际执行步数也为25；失败发生在飞机循迹操作中。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.8,
+    "log_step": 135,
+    "evidence_zh": "观察：视频0.80–1.00 s，实际飞机保持在红色手下方的桌面附近，朝向和位置没有随分离的绿色参考显示共同变化，未形成要求的空中飞行过程。log step 135最终生成会话的位移误差由控制步20的0.0597 m增至控制步25的0.25538 m，与视频后段的空间偏离相符。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 134,
+    "evidence_zh": "推断：log step 134的最后交付脚本以参考位姿生成前臂逆解和腕部、手指肌肉控制，并增加屈肌激活来合手；log step 135执行该脚本时，飞机仅小幅升高（控制步24相对初态14.8 mm），位移误差却迅速增加并提前终止。视频0.60–1.00 s也显示合手后飞机只在桌面附近移动，支持控制未能产生要求的持物循迹运动；缺少本次逐步关节与接触数据，不能细分接触设计、逆解或反馈控制。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "未提供状态日志和成功示范；视频仅640×240且叠加参考手形，无法独立测量厘米级误差、接触力和各指承载。输入reward未报告DAG列出的四项原生循迹数值，故native_metrics均为null；log step 135中的误差和计数属于最后生成会话，未冒充评分重放的原生数值。最后控制脚本的日志中间部分被截断，也无法据此定位具体控制子类。"
+ },
+ "summary_zh": "单手合拢夹住飞机并带动其在桌面附近移动，但未完成飞机循迹飞行，节点最终0/1。视频0.80–1.00 s显示飞机与参考标记持续分离，最后保存的生成会话在控制步25提前终止，评分器同样只执行25步；证据支持位姿不准及一般控制失败。缺少状态日志、接触力与评分重放的精确循迹指标，不能细分控制根因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "单节点未达成、断点在该节点、位姿不准与停在“控制”的根因均有视频和评分器25步终止支持；时钟未把控制步换算成视频时间，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:58:09Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--p6hl228/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--p6hl228/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "02ccee890d7f4fd0ba8158f56323dc87d2b7ad2212f6fee6aebd7788d13d99c6",
+   "video_sha256": "061e8136c533839b105b25408e7ae7016228e86a7929518372e64e95e55bf6ff",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--pawfrco.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--pawfrco.json
new file mode 100644
index 0000000000000000000000000000000000000000..eadbcd927ce3ecd7ca46a2975627f6d576be5c6b
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--pawfrco.json
@@ -0,0 +1,199 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-privileged--pawfrco",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 38,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 141,
+   "agent_resets": 5034,
+   "agent_sim_steps": 5115,
+   "agent_records": 103,
+   "agent_replays": 3,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.515,
+   "peak_processes": 17,
+   "peak_disk_gb": 0.243,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-privileged--pawfrco/replay.mp4",
+  "duration_s": 1.56,
+  "fps": 25.0,
+  "frames": 39,
+  "sha256": "e1e646924f83c4ced3ac1a174be7b8faea34397ba23461185577b903b673297c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.68 s，MyoHand单手向桌面上的白色玩具飞机合拢，飞机随后随手离开原来的桌面位置；视频 1.20–1.52 s，飞机仍夹在弯曲的手指之间并随手转动。"
+     },
+     {
+      "t": 1.36,
+      "log_step": 93,
+      "text_zh": "观察：视频 1.20–1.52 s，白色飞机仍在手中，但绿色参考标记逐渐向画面左下方远离飞机；log step 93 对最终提交文件进行两次独立重放，均只执行38/99个控制步，原因均为飞机距参考或手腕超过0.25米，终态哈希相同；评分器reward亦报告sim_steps=38、deterministic=1、task_failure=1。完整99步循迹未完成，不能因开局误差小而记为完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.56–0.68 s及1.20–1.52 s，MyoHand单手的拇指和弯曲手指围住飞机的翼根、机身后部附近，飞机随手抬起、转动；推断：由合拢包夹及持物共同行动判断为抓握，具体接触点没有状态日志可精确核对。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 1.28,
+   "log_step": 93,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 1.20–1.52 s，机器人已把飞机带离桌面并继续持握，但飞机与运动中的绿色参考标记明显分离，且差距持续扩大至片尾；log step 93 验证的最终提交在控制步38提前终止，未走完99步。推断：失败发生在持物循迹过程中，视频约1.28 s为开始明显拉开差距的导航锚点，并非从agent时间换算。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 1.28,
+    "log_step": 93,
+    "evidence_zh": "观察：视频 1.20 s绿色参考标记还在白色飞机附近，视频 1.28–1.52 s标记向画面左下方持续移走，机器人持握的飞机仍停留在手旁，未跟上参考的移动；片尾没有看见飞机脱手。log step 93 的最终文件重放因飞机距参考或手腕超过0.25米提前终止。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：agent脚本的运动控制未能带着已抓住的飞机跟上参考。log step 79 的脚本在抓取后将参考臂腕增量加到抓取姿态上，并按物体误差反复修正；log step 88 的修正试验仍在控制步38左右终止，log step 89 保存并重放的最终动作仅26步共同达标、控制步38终止，log step 93 两次独立重放复现该结果，与视频 1.28–1.52 s持物滞后的现象一致。会话诊断log step 83–84还显示指令与实际臂部运动存在偏差，但它们不是评分状态日志，因此不据此细分反馈控制、逆解或参数问题。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功参考视频；右视角在视频 0.72–1.16 s被桌面遮挡，左视角仍能看到手与飞机，但不能精确还原各指接触力。reward未报告track_pos_err、track_rot_err、track_ok_fraction和track_max，故四项原生指标留null；无法定位首次越过0.05米的精确视频时刻，断点仅用可见分离作粗略锚点。log step 95 关于载荷、抓法后续不可达的解释是agent自述，会话诊断也未排除其他合法策略，不能据此认定任务无解或进一步确定控制子类。"
+ },
+ "summary_zh": "机器人单手夹住并抬起玩具飞机，但唯一的“飞机循迹飞行”节点未完成，最终为0/1。视频 1.28–1.52 s可见飞机仍在手中而参考标记持续远离；log step 93确认最终文件两次重放均在控制步38/99提前终止，支持运动控制未能跟上动态参考。缺少状态日志和评分器原始循迹指标，不能精确核对误差越限时刻或细分控制根因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未完成(0/1)、单手抓握飞机后参考标记远离、最终文件重放控制步38/99终止，均与视频及log step 93一致；断点、动态时机现象和控制根因表述克制，未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:54:31Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--pawfrco/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--pawfrco/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "41a5fd54d7076e6e653568be0a60561768843429cb60b32f2233aa714a9633e5",
+   "video_sha256": "e1e646924f83c4ced3ac1a174be7b8faea34397ba23461185577b903b673297c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--wnelj7s.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--wnelj7s.json
new file mode 100644
index 0000000000000000000000000000000000000000..b189cf6be84d57f8f9fd3f0c02c9a95471a32ae3
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-privileged--wnelj7s.json
@@ -0,0 +1,206 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-privileged--wnelj7s",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 37,
+   "sim_steps": 37,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 102,
+   "agent_resets": 400,
+   "agent_sim_steps": 12963,
+   "agent_records": 0,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.435,
+   "peak_processes": 21,
+   "peak_disk_gb": 0.251,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-privileged--wnelj7s/replay.mp4",
+  "duration_s": 1.52,
+  "fps": 25.0,
+  "frames": 38,
+  "sha256": "596f6c1af80333ffe478b3e75ff6973023a5d07761cdc37c5315bc6d7561fad6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 双视角显示玩具飞机在桌面上，MyoHand 单手手指位于飞机上侧。逐帧检查视频 0.00–1.48 s，飞机只有少量桌面内位移，没有被手带起或在空中飞行。"
+     },
+     {
+      "t": 1.08,
+      "log_step": null,
+      "text_zh": "观察：视频 1.00–1.48 s，手指继续屈曲并贴压飞机上侧，飞机仍贴着桌面；未出现包住机身并带起飞机的夹持。"
+     },
+     {
+      "t": null,
+      "log_step": 171,
+      "text_zh": "观察：log step 162 的生成命令保存了 37×45 的动作数组；log step 171 对该提交文件进行两次重放，均在 37 个控制步后返回 success=false，原因是飞机距参考或手腕超过 0.25 m，终态哈希均为 89627a70296b910b。该结果与视频中飞机未起飞一致，也未走完 DAG 要求的 99 个控制步。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.80–1.48 s，MyoHand 单手弯曲的手指贴压玩具飞机上侧及尾部附近；飞机仍由桌面承托，未观察到对向包夹并抬升飞机的抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 1.08,
+   "log_step": 171,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 1.00–1.48 s，MyoHand 正在飞机上方屈指，飞机始终留在桌面，没有被带起；这一尝试直到片尾都未修复。log step 171 的提交文件重放在 37 个控制步后因飞机距参考或手腕超过 0.25 m 提前结束。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 1.08,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.80–1.48 s，MyoHand 的手指在飞机上侧弯曲、压触，飞机仍在桌面；没有形成使飞机离桌并随手运动的夹持。这里缺少的是承载飞机所需的抓握接触，未观察到先拿稳再掉落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": null,
+    "log_step": 161,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 161 的提交生成脚本以参考手关节位置误差和关节速度计算力矩，再用非负最小二乘分配肌肉激励；脚本未依据飞机运动或接触状态调整夹持。视频 0.80–1.48 s 显示执行时手指只有上侧压触，未使飞机离桌，log step 171 又确认该提交可确定性重现同一失败，支持将未形成有效夹持归于控制；不足以确定是具体接触布局、肌力分配还是关节跟踪所致。"
+   }
+  ],
+  "gap_zh": "没有状态日志和成功参考视频；现有双视角可判定飞机未被带起，但不能精确恢复每根手指的接触力和关节跟踪误差。评分器 reward 字段及提交文件重放结果未报告 track_pos_err、track_rot_err、track_ok_fraction、track_max，故这四项数值留空；会话中其他测试的数值不作为提交轨迹的指标。控制根因的具体子类仍不确定。"
+ },
+ "summary_zh": "观察：逐帧检查视频 0.00–1.48 s，MyoHand 手指在飞机上侧屈曲、压触，但飞机始终留在桌面，“飞机循迹飞行”未完成，最终进度为 0/1。推断：尝试在建立可带起飞机的夹持时失败，归为控制问题；log step 171 确认已提交的 37 步轨迹两次确定性重放均提前失败。无状态日志、成功示范或提交轨迹的精确循迹指标，尚不能细分控制原因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、没建立接触、控制根因均与视频和log相符；未发现需改动的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:59:18Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--wnelj7s/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-privileged--wnelj7s/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "4a51487f7f963963b6309ccb6b306da5c15c444e8d549f804f3d07ad7a1ed19c",
+   "video_sha256": "596f6c1af80333ffe478b3e75ff6973023a5d07761cdc37c5315bc6d7561fad6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--bnj3e9h.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--bnj3e9h.json
new file mode 100644
index 0000000000000000000000000000000000000000..ab41316142f24006cb2edda0cf2e8779df98e296
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--bnj3e9h.json
@@ -0,0 +1,221 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-standard--bnj3e9h",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 38,
+   "sim_steps": 38,
+   "physics_steps": 380,
+   "steps": 85,
+   "calls": 8,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 85,
+   "elapsed_s": 2207.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.088,
+   "peak_processes": 17,
+   "peak_disk_gb": 0.444,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-standard--bnj3e9h/replay.mp4",
+  "duration_s": 1.56,
+  "fps": 25.0,
+  "frames": 39,
+  "sha256": "1b24d42ff48ffdef8c204c1e3cb0fd11aa42df6ebf5ef4ab68f8931aca546135"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频0.00–1.52 s的全部39帧；单只MyoHand的拇指与弯曲手指围住白色飞机，飞机随手移动并转向，因而本步确实开始尝试。"
+     },
+     {
+      "t": 1.4,
+      "log_step": null,
+      "text_zh": "观察：视频1.28–1.52 s，绿色参考标记继续向画面左下方移走，手及所带飞机未同步跟上，末段明显分离；未看到完整循迹完成。"
+     },
+     {
+      "t": null,
+      "log_step": 63,
+      "text_zh": "观察：log step 63的服务返回episode=3、control_steps=38、done=True，原因为飞机距参考或手腕超过0.25米；inputs/run.json评分字段n_actions=38、sim_steps=38、success=0。log step 58属于此前的episode=2，不用于本局进度。"
+     },
+     {
+      "t": null,
+      "log_step": 63,
+      "text_zh": "推断：最后评分局仅执行38个控制步即原生终止，未满足DAG要求的完整99步，因此该整段运动事件始终未完成；这不是仅由零分推断未尝试。视频锚点直接取所查帧，不将agent工作时间或20毫秒控制步直接当作视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频0.00–1.28 s，MyoHand单手拇指与弯曲手指从相对两侧围夹飞机，随手移动及转向；并非双手协作。视频局部分辨率和遮挡不足以逐指确认整个末段的接触维持情况。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_rot_err",
+   "value": null,
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok_fraction",
+   "value": null,
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 1.28,
+   "log_step": 63,
+   "where": "操作中",
+   "evidence_zh": "观察：视频1.28–1.52 s，机器人仍在带动飞机尝试循迹，绿色参考标记与手中飞机逐渐拉开，未恢复跟随；log step 63确认最后评分局episode=3在control_steps=38因飞机距参考或手腕超过0.25米而终止。推断：失败发生在飞机循迹飞行的操作中；视频1.28 s为明显失配的粗略导航锚点，不是精确阈值越界时刻。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "动态时机",
+    "t": 1.28,
+    "log_step": 63,
+    "evidence_zh": "观察：视频1.12–1.28 s绿色参考标记仍在飞机附近，视频1.36–1.52 s标记继续移向左下而机器人手及飞机的运动未同步，间隔明显扩大；log step 63服务返回距离超限终止。推断：这是对随时间变化的参考运动未跟上，表现为动态时机失配。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "观察：log step 47生成的执行脚本将离线优化的整批执行器指令以dt=0.02提交给服务，没有依据执行中的物体误差进行修正；log step 63实际执行这一路径并在38个控制步后失败。log step 51的离线抓持测试中，后段物体升高约0.048米、腕升高约0.110米且相对腕位置偏移约0.043米；这些是会话诊断而非评分局状态。推断：结合视频1.28–1.52 s的跟随失配，所交付的肌肉与前臂控制未能可靠带动飞机跟上参考；证据不能分清抓持与轨迹跟踪的贡献，根因停在控制层。"
+   }
+  ],
+  "gap_zh": "无状态日志、成功示范或评分局逐控制步的未舍入误差；评分字段未报告track_pos_err、track_rot_err、track_ok_fraction、track_max，四项指标均填null。双视角存在过曝、遮挡及相机运动，可辨认尝试与后段跟随分离，但不能精确确定失配阈值的首次越界、逐指接触或原生终止究竟由参考距离还是腕距触发。agent在log step 65自述抓持滑移及竖直执行器能力不足，不能据此证实环境不可解，也不将其离线诊断当作本局精确状态。"
+ },
+ "summary_zh": "最后评分局中MyoHand抓住并带动飞机移动、转向，但视频后段未跟上参考，完整的“飞机循迹飞行”节点未达成（终态0/1）。log step 63确认该局在38/99个控制步时触发原生距离终止，现象为动态时机失配，根因保守归为控制。缺少状态日志和原生误差数值，无法确定抓持滑移与前臂跟踪各自的影响或精确阈值越界时刻。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与证据一致：最后评分局（episode 3，log id 63，38/99 步原生终止）里手抓住并带动飞机，后段未跟上参考；节点未达成、断点、动态时机现象与保守的控制根因均有支撑，时钟未把agent时间转换为视频时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:52:45Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--bnj3e9h/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--bnj3e9h/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a9cbb3654ab5b395b6a87ffb13ac17aaac055a784523d1e0ed8c6582d6c61236",
+   "video_sha256": "1b24d42ff48ffdef8c204c1e3cb0fd11aa42df6ebf5ef4ab68f8931aca546135",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--edde464.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--edde464.json
new file mode 100644
index 0000000000000000000000000000000000000000..fbc6489d056bb21521b9493cb9fbe410aa6946c3
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--edde464.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-standard--edde464",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 290,
+   "calls": 81,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 290,
+   "elapsed_s": 1184.7,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.337,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.84,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-standard--edde464/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "40a18b6d3b6273d9f66c948cc43020a261cc156f61d044f0f54f21069d8ad894"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频0.00–0.76 s的全部20帧；开局白色飞机在桌面上，MyoHand指端位于机身上方。视频0.16–0.32 s手指卷曲，视频0.56–0.76 s右侧视图中空手与飞机明显分离，未见飞机被夹持并带起完成空中循迹。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 80,
+      "text_zh": "观察：log step 80明确运行最后的episode 11，服务最终状态为control_steps=19、done=true，reason为物体距参考或手腕超过0.25 m；评分器n_actions=19、live_matches_replay=1、success=0。该局未走完节点要求的99个控制步；视频中的空手闭合与此局的抓取失败一致。未将日志仿真0.38 s换算为视频时刻。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频0.00–0.76 s中，MyoHand单手指端在飞机上方闭合后空手移离，未见形成对向夹持。低分辨率及初始指端重叠使瞬时轻触及其方式无法确认，故实际接触记未知，不将抓取指令记为已发生的抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_rot_err",
+   "value": null,
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok_fraction",
+   "value": null,
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.32,
+   "log_step": 80,
+   "where": "操作中",
+   "evidence_zh": "观察：视频0.16–0.32 s，MyoHand手指逐渐闭合，但白色飞机仍在手下方的桌面上；视频0.56–0.76 s手中仍无飞机，未恢复抓取。log step 80是最后一局，服务在control_steps=19时以物体距参考或手腕超过0.25 m终止，该局没有完成飞机循迹飞行。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.32,
+    "log_step": 80,
+    "evidence_zh": "观察：视频0.00–0.76 s中，MyoHand手指从飞机上方的张开姿态卷成空握，飞机未随手离开桌面；视频0.32 s已能看到指端与机身间的间隙，后续逐帧至视频0.76 s始终未见包夹。所需的承载抓握没有建立，无证据表明曾抓住后滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log steps 78–79的最终控制器用固定的手指肌肉激励闭合，并按初始手与物体的相对变换计算掌部目标；log step 80中，控制器在尚未抓起飞机时继续执行掌部运动，例如仿真0.20 s的q前两项为[-0.042,-0.076]而q_cmd为[-0.158,-0.038]。结合视频0.16–0.76 s手指在空中闭合并与飞机分离，说明抓取及手部运动控制未实现所需接合；不足以进一步区分接触设计、逆解或动态跟踪，故停在控制层级。"
+   }
+  ],
+  "gap_zh": "观察：inputs/提供完整视频和运行日志，但无状态日志、成功示范或评分器的track_pos_err、track_rot_err、track_ok_fraction、track_max数值，四项原生指标均留null。初始指端重叠且视频分辨率低，无法确认短暂轻触或具体接触方式；能确认未建立承载抓握。视频20帧共0.80 s，而log step 80报告19个20毫秒控制步、仿真0.38 s；没有经核对的两者时间映射，所有导航时刻直接取视频帧，根因不细分，脚本打印的plane估计不作为飞机真实位姿。"
+ },
+ "summary_zh": "最后评分局中，MyoHand在飞机上方闭合手指后空手移离，未抓起飞机，飞机循迹飞行节点未完成（0/1）。断点在该步操作中，现象为没建立接触；最终控制器的抓取及掌部运动未实现所需接合，根因保留为控制，服务在19个控制步后原生终止。无状态日志、成功示范及精确循迹指标，初始轻触方式和控制子类无法确定，视频与日志仿真时长也未建立映射。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、没建立接触与控制根因均与视频帧一致；时间均取视频帧，未换算日志时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:11:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--edde464/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--edde464/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "88e70d74efea6cb6370baf4ed068c2c9214d5b27dfad16fe4fc6180021f1e029",
+   "video_sha256": "40a18b6d3b6273d9f66c948cc43020a261cc156f61d044f0f54f21069d8ad894",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--emurq9d.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--emurq9d.json
new file mode 100644
index 0000000000000000000000000000000000000000..064c63e0c896da343416db4c9688e5f00de50c7e
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--emurq9d.json
@@ -0,0 +1,219 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-standard--emurq9d",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 37,
+   "sim_steps": 37,
+   "physics_steps": 370,
+   "steps": 214,
+   "calls": 196,
+   "episodes": 6,
+   "resets": 5,
+   "reset_budget": 10,
+   "sim_steps_total": 214,
+   "elapsed_s": 3566.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.253,
+   "peak_processes": 15,
+   "peak_disk_gb": 1.761,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-standard--emurq9d/replay.mp4",
+  "duration_s": 1.52,
+  "fps": 25.0,
+  "frames": 38,
+  "sha256": "6ee0c58dcf0fe0cbfca294b5ce283004044e647bde02cde5ccc2b8377b051bba"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 两个视角显示玩具飞机在桌面，MyoHand位于飞机一侧上方；初态不代表整段循迹飞行已完成。"
+     },
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–1.48 s；MyoHand手指逐渐收拢，视频 0.72–0.96 s 的闭合手指仍在机身一侧上方，未包住机身；视频 1.12–1.48 s 手指贴近机翼，但飞机仍贴着桌面，没有被手带起飞行。"
+     },
+     {
+      "t": 1.48,
+      "log_step": 239,
+      "text_zh": "观察：视频 1.48 s 飞机仍在桌面；log step 236 重新运行修正后的控制器，log step 237 返回执行37个控制步后原生终止，原因是物体距参考或手腕超过0.25 m，log step 239 的服务状态确认episode=6、control_steps=37、done=True。评分器reward.graded_by_live=1、sim_steps=37，判定最后一局失败；未走完所需99步。日志中的控制步未换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "推断：视频 1.12–1.48 s，MyoHand单手收拢的手指贴在飞机一侧机翼上方，按贴压归类；飞机保持桌面支承，未见对向夹住机身或从下方托起，不记抓握。无接触对或力传感记录，具体触点无法精确确认。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.8,
+   "log_step": 237,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80 s，MyoHand手指已经收拢在飞机一侧上方，飞机仍留在桌面；随后视频 1.12–1.48 s 也未被抓起。log step 237 记录最后一局在37个控制步后因物体距参考或手腕超过0.25 m终止，log step 239 确认这是第6局最终服务状态；飞机循迹飞行未完成，前面的抓取失败始终未修复。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.72–0.96 s，MyoHand手指在飞机机身一侧上方闭合，飞机没有进入手指围成的夹持区域；视频 1.12–1.48 s 虽有贴压机翼附近的动作，飞机始终未随手离开桌面。推断：循迹所需的承载抓握未建立，不是先抓起后滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.8,
+    "log_step": 236,
+    "evidence_zh": "推断：视频 0.72–1.48 s 显示手指在飞机一侧上方闭合，后段贴压机翼附近，未在机身两侧形成可承载的夹持。log step 231 的控制代码先调整前臂、再以同组屈指肌逐渐收拢，随后直接切换参考轨迹；log step 236 仅删除越界索引后重置并执行此策略，log step 237 记录正常执行至原生物体距离终止。结合可见的空夹和飞机未抬起，所用接触布局未抓住飞机；没有在本局验证成功的替代握法。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "无状态日志、接触对或接触力记录，贴压方式由可见接近与闭合几何推断，不能精确确定各指触点。没有成功示范；评分器reward未报告track_pos_err、track_rot_err、track_ok_fraction、track_max，四项原生指标均留null，不能从视频量出精确误差。agent在log step 240 将问题归于物体位姿估计，但现有证据不足以独立确认错误估计如何导致本局动作，故不另标感知根因。"
+ },
+ "summary_zh": "评分的第6局中，MyoHand收指并在飞机一侧贴压，但未建立能带起飞机的抓握，飞机始终留在桌面，飞机循迹飞行完成0/1。断点位于抓取过程，推断闭合位置与接触布局不适配；服务在37个控制步后因物体距参考或手腕超过0.25 m终止。无状态日志和精确循迹指标，具体触点及替代握法可行性仍不确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频逐帧显示手指在飞机一侧收拢、飞机始终留在桌面、绿色参考目标逐渐远离，节点未达成、断点与现象（没建立接触）、根因均有证据支持，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:52:40Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--emurq9d/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--emurq9d/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f4a69ef5ec16d4a1bc9c262361a3269eb073202c6220e117a86c6f08d423a77d",
+   "video_sha256": "6ee0c58dcf0fe0cbfca294b5ce283004044e647bde02cde5ccc2b8377b051bba",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--erawpc2.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--erawpc2.json
new file mode 100644
index 0000000000000000000000000000000000000000..c1ef8648bba5ed65b331d075e0721ecdf9f94c70
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--erawpc2.json
@@ -0,0 +1,215 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-standard--erawpc2",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 11,
+   "sim_steps": 11,
+   "physics_steps": 110,
+   "steps": 11,
+   "calls": 2,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 11,
+   "elapsed_s": 3590.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.702,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.327,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-standard--erawpc2/replay.mp4",
+  "duration_s": 0.48,
+  "fps": 25.0,
+  "frames": 12,
+  "sha256": "8297f8d0a02e2bfdc3e040199a5f2ff4001a875c8f5c9cc6a3a73427d408c7f7"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s，MyoHand单手位于桌面附近的玩具飞机上方，飞机尚未被带起。"
+     },
+     {
+      "t": 0.28,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频 0.00–0.44 s 的全部12帧；手指屈伸靠近飞机，视频 0.24–0.44 s 左侧视图中飞机向手的左侧移开，末帧仍未见飞机被手稳定抓持并带起飞行。"
+     },
+     {
+      "t": 0.44,
+      "log_step": 153,
+      "text_zh": "观察：log step 153 中，机器人服务在提交前报告control_steps=0；提交99行执行器命令后只执行11步，返回done=True及“failure: object more than 0.25 m from the reference (or from the wrist)”。评分器字段n_actions=11、task_failure=1、live_matches_replay=1；评分局未走完节点所需的99步。此处视频时间仅定位末帧，不把服务的仿真 0.22 s 换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "推断：视频 0.12–0.44 s，MyoHand单手手指在飞机上方及侧缘屈伸，飞机随后向手外侧移开，表现为手指拨动飞机；未见相向夹持后稳定带起，故不记抓握。具体接触手指与短暂接触的先后无法从低分辨率画面确定。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.24,
+   "log_step": 153,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.24–0.44 s，机器人手指在飞机附近屈伸，而飞机向手外侧移开，没有被稳定抓持并带起。log step 153 的机器人服务在执行11步后因飞机距参考或手腕过远而终止；该抓取及循迹尝试没有后续修复。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.24,
+    "log_step": 153,
+    "evidence_zh": "推断：视频 0.00–0.44 s 的逐帧画面中，机器人手指虽靠近并拨动飞机，但未形成能让飞机随手共同运动的稳定抓握；飞机向手外侧移开，未见先被持稳再掉落。结合log step 153 服务在11步后以飞机距参考或手腕过远终止，判断抬起所需的抓握没有建立；这里指所需抓握未建立，并非完全没有物理触碰。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 153 的agent脚本在本地MuJoCo中计算99行肌肉及前臂执行器命令，再一次性提交机器人服务；同一步的本地测试已显示控制步30飞机质心高度0.0182米、参考高度0.0835米，控制步50仍为0.0182米而参考为0.4467米，未实现有效带起。实际评分视频 0.00–0.44 s 同样未形成稳定抓持，服务只执行11步即终止，支持执行器控制未实现抓取这一宽泛归因。本地测试仅作为根因旁证，不用于评分节点进度；现有证据不能区分接触设计、模型差异或跟踪控制的具体责任。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功示范；视频分辨率与局部遮挡限制了具体接触手指、接触力及瞬时夹持的判定，推/扫是依据手指动作与飞机移开的上下文推断。log step 153 服务返回的仿真时长为0.22秒，视频时长为0.48秒，缺少可核对的逐帧控制步映射，未按控制步换算视频时间。评分器reward未报告track_pos_err、track_rot_err、track_ok_fraction、track_max，服务输出亦有截断，四项数值均保留null；不能用agent本地仿真的误差替代实际评分指标，也不能进一步确认控制根因子类。"
+ },
+ "summary_zh": "评分尝试仅执行11个控制步，飞机循迹飞行节点未完成，最终进度0/1。视频 0.00–0.44 s 显示手指未建立能稳定带起飞机的抓握，飞机向手外侧移开；log step 153 的机器人服务因飞机距参考或手腕过远而终止，根因证据支持宽泛的控制问题。缺少状态日志、精确循迹指标及仿真到视频的可靠时间映射，无法细分接触和控制机制。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "单节点未达成、最终0/1、断点在操作中均与视频和评分器字段一致：视频0.24–0.44 s飞机被手指拨开并飞离，服务在11步后因飞机距参考或手腕超过0.25 m终止；没建立接触符合定义，控制根因标为宽泛归因且已声明本地仿真只作旁证。未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:54:42Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--erawpc2/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--erawpc2/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "82e4bf0b3154e315b00831763e77229ec3771bb644992725811546e69969c453",
+   "video_sha256": "8297f8d0a02e2bfdc3e040199a5f2ff4001a875c8f5c9cc6a3a73427d408c7f7",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--hyul5i7.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--hyul5i7.json
new file mode 100644
index 0000000000000000000000000000000000000000..90393e3478eb411e19154325c07bc7157ef7bd68
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--hyul5i7.json
@@ -0,0 +1,221 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-standard--hyul5i7",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 20,
+   "sim_steps": 20,
+   "physics_steps": 200,
+   "steps": 293,
+   "calls": 113,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 293,
+   "elapsed_s": 2767.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.164,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.523,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-standard--hyul5i7/replay.mp4",
+  "duration_s": 0.84,
+  "fps": 25.0,
+  "frames": 21,
+  "sha256": "46e296878c49b02d641a2eab8dd8664a121f209f495d076d41c20a0359b5157e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 两个视角显示玩具飞机在桌面附近，MyoHand手指位于飞机上方；右侧近景未见手指围住机身。"
+     },
+     {
+      "t": 0.16,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.80 s 的全部21帧，视频 0.16 s 起手指与飞机的间隙增大，随后手部上移；飞机留在桌面附近，至视频 0.80 s 仍未随手抬起或进行空中循迹。"
+     },
+     {
+      "t": null,
+      "log_step": 368,
+      "text_zh": "观察：log step 368 中agent执行最终脚本，服务从第10局重置为第11局，接收99步提交但仅执行20个控制步即done=True，原因为飞机距参考或手腕超过0.25米；log step 373确认同一终态。评分字段graded_by_live=1、n_actions=20、success=0。"
+     },
+     {
+      "t": null,
+      "log_step": 368,
+      "text_zh": "推断：结合视频中飞机始终未被带起和log step 368的第20控制步提前终止，本局没有完成节点要求的99步参考运动；不能把agent在log step 375自述的运动学误差达标当作飞机实际循迹成功。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–0.80 s 中MyoHand单手在飞机上方屈指并上移，未见对向夹紧机身或带起飞机。开头指尖与飞机尾部局部投影接近，缺少接触对记录，不能确认是否有瞬时碰触及其方式；不据脚本的抓握注释认定实际抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.16,
+   "log_step": 368,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.16 s 手部开始明显向上离开飞机，视频 0.32–0.80 s 飞机仍在桌面附近，没有随手升起；log step 368显示最终第11局在第20控制步触发原生距离终止，未完成99步运动。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.16,
+    "log_step": 368,
+    "evidence_zh": "观察：视频 0.00–0.16 s 未见手指合拢夹住飞机；视频 0.16–0.80 s 手部屈指上移，飞机留在桌面附近。推断：循迹所需的承载抓握未建立，未见已经持有飞机后再滑脱的过程；log step 368的原生距离终止与手、飞机分离的画面一致。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "观察：log step 363的最终脚本将target_q[:6]-q0[:6]写入六个根执行器命令，并对指定肌肉使用整段固定激活，一次提交99步；log step 12说明这六个执行器接收关节位置目标。log step 368确认执行该脚本的第11局仅运行20步即终止。推断：结合视频 0.16–0.80 s 的空手上移与飞机留在桌面附近，这组控制命令未能在移动前建立并维持承载夹持；现有证据不能区分接触几何、肌肉激活和根执行器目标换算的各自贡献，故停在控制大类。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有状态日志、成功示范或最终第11局的逐步位置误差、转动误差、达标比例与最大误差，四项原生指标均未报告；开头指尖与飞机局部投影接近，无法确认瞬时碰触，但完整视频可确认没有抓住并带起飞机。log step 368记录20个控制步、命令dt=0.02 s，对应仿真时长0.40 s，而视频时长0.84 s，未建立两者映射，所有视频锚点均直接来自已检查的帧。"
+ },
+ "summary_zh": "观察：最终评分为log step 368重置后的第11局，视频中MyoHand手部屈指后上移，飞机仍留在桌面附近，飞机循迹飞行节点未完成，终态0/1。推断：失败断在操作中的承载抓握未建立，归入控制；服务在第20控制步因飞机距参考或手腕超过0.25米终止，而非执行完99步。缺少状态日志及逐步循迹误差，不能细分控制根因，也不能确认开头是否有短暂碰触。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "逐帧核对视频：手指上移，飞机始终留在桌面，节点未达成、断点与现象与证据一致，时间均为视频钟，未转换agent时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:57:08Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--hyul5i7/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--hyul5i7/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "32a2f7dccc4d29e26635f37643a9e8e11d9e4b83561637eb5275b25cddbe4d57",
+   "video_sha256": "46e296878c49b02d641a2eab8dd8664a121f209f495d076d41c20a0359b5157e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--yknseap.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--yknseap.json
new file mode 100644
index 0000000000000000000000000000000000000000..ddc78f72301c6efd6660cb8b94700745f7b47393
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--yknseap.json
@@ -0,0 +1,217 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-standard--yknseap",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 37,
+   "sim_steps": 37,
+   "physics_steps": 370,
+   "steps": 392,
+   "calls": 421,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 392,
+   "elapsed_s": 3485.6,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.668,
+   "peak_processes": 17,
+   "peak_disk_gb": 1.387,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-standard--yknseap/replay.mp4",
+  "duration_s": 1.52,
+  "fps": 25.0,
+  "frames": 38,
+  "sha256": "d0849eaa8c3d3497cebeaba27722f681609b0ca639f4c21369a07db9b730aff0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.48,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频 0.00–1.48 s 的全部38帧；机器人单手弯曲手指并移动，右侧视角中飞机仍在桌面，视频 0.48–0.96 s 手与飞机之间的间隔增大，未见手指夹持飞机并将其带起。"
+     },
+     {
+      "t": 1.48,
+      "log_step": 143,
+      "text_zh": "观察：视频 1.20–1.48 s 飞机仍在桌面，手指在其上方空置；log step 143 的机器人服务状态为 episode=11、control_steps=37、done=true，原因为飞机距参考或手腕超过0.25米。该局未走完99个控制步，与视频中未完成抬起和循迹一致；这里只并列两种证据，不把控制步或agent工作时间换成视频时间。"
+     },
+     {
+      "t": null,
+      "log_step": 137,
+      "text_zh": "观察：log step 137 调用 final_run(stop=24, reset=True) 启动最后一局；log step 140 的服务状态确认 episode=11、resets_left=0；log step 142 继续该局，log step 144 再次确认其在control_steps=37失败。inputs/run.json 的reward记录 graded_by_live=1、n_actions=37、episodes=11，故此前重置掉的试验不计为本次节点进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–1.48 s，MyoHand单手在飞机尾部附近弯曲手指、尝试捏取，未见包围或夹住飞机，不能记为已发生的抓握。开局指尖与尾部投影接近，画面分辨率不足以排除短暂擦碰，且无接触对日志，因此实际接触方式记未知。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.6,
+   "log_step": 141,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.48–0.96 s，机器人手指弯曲后随前臂移动，飞机仍留在桌面，没有形成可带起飞机的夹持；到视频 1.48 s 仍未修复。log step 141 检查的最终局控制步24显示实际关节位置偏离计划；log step 143 的服务状态随后在控制步37以飞机距参考或手腕超过0.25米终止。断点取正在执行的飞机循迹飞行，视频 0.6 s 仅作抓取未建立的粗略导航锚点。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.6,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–1.48 s，机器人单手在飞机附近弯曲手指，但飞机没有随手离开桌面；视频 0.60、0.96、1.20及1.48 s 的右侧视角显示手指与机体分离，未建立抬起所需的夹持接触。未见曾抓住飞机后再掉落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 141,
+    "evidence_zh": "推断：最终局的执行未把手送到计划的捏取位置。log step 141 对比该局控制步24的计划与本体感觉关节位置，第二个平移分量计划为-0.03078米、实际为-0.08147米；该脚本用手模型计算的实际拇指/食指尖高度约0.0969/0.1099米，计划约0.0626/0.0656米。这些模型计算不是物体状态真值，但与视频 0.48–1.48 s 手指空置、飞机留桌一致。log step 142 的控制脚本随后尝试修正捏点，log step 143 仍显示未跟上其目标。证据支持控制执行偏差，尚不足以确定是逆解、反馈结构或具体参数所致。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "无模拟器状态日志或接触对记录；开局指尖与机尾邻近处不足以确认是否有短暂擦碰，因此contact_actual的接触方式为未知，但全片足以确认没有建立有效夹持。评分器reward未报告track_pos_err、track_rot_err、track_ok_fraction或track_max的原始值，四项原生指标均置null；无成功示范，不能进一步验证具体控制子类或替代抓法的可行性。视频时钟直接按25 fps取帧，未把50 Hz控制步或agent工作时间直接映射为视频时间。"
+ },
+ "summary_zh": "最终评分局中，机器人单手尝试捏取飞机，但飞机始终留在桌面，没有完成抬起及参考循迹，节点终值为0/1。断点为飞机循迹飞行操作中未建立有效夹持；最终局日志显示控制执行偏离计划，服务在控制步37因飞机距参考或手腕超限而终止。无状态接触记录和原始循迹误差，短暂擦碰及更细控制根因无法确认。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点、没建立接触与控制根因均与视频（1.5 s 内飞机始终在桌面、手指空置）及最终局日志一致；时钟未换算，未发现需修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:59:10Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--yknseap/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--yknseap/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6562e3673eed1adfe66b6e9b6273768430375d6b1bccad4f1ffd7cd3b35ae97c",
+   "video_sha256": "d0849eaa8c3d3497cebeaba27722f681609b0ca639f4c21369a07db9b730aff0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--yyrojfo.json b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--yyrojfo.json
new file mode 100644
index 0000000000000000000000000000000000000000..717bbb45cdf0ec44beec5cc3d123b28a777e5872
--- /dev/null
+++ b/phase2/myosuite/dex_airplane_fly/trajectories/myosuite-dex-airplane-fly-i00-standard--yyrojfo.json
@@ -0,0 +1,216 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-airplane-fly-i00-standard--yyrojfo",
+ "taskKey": "myosuite/dex_airplane_fly",
+ "dag_sha256": "211c82c127093021391297ba04609fe900a6c8513c288e6a7fb1169216a9dbdd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 35,
+   "sim_steps": 35,
+   "physics_steps": 350,
+   "steps": 361,
+   "calls": 27,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 368,
+   "elapsed_s": 1336.9,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.538,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.581,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-airplane-fly-i00-standard--yyrojfo/replay.mp4",
+  "duration_s": 1.44,
+  "fps": 25.0,
+  "frames": 36,
+  "sha256": "f8ef8c2ed2468b5ef53c368587dc41f2e2061e795d2413c58fbd044d7232b121"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "airplane_reference_motion_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的右侧外部视角显示飞机在桌面，MyoHand手指位于飞机上方；逐帧检查视频 0.00–1.40 s，飞机未随手升起。"
+     },
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：视频 0.32–0.60 s 手指逐渐收拢，手与前臂向上离开飞机；视频 0.84–1.40 s 右侧视角中手已明显空悬，飞机仍在桌面，未出现持机飞行或转向。"
+     },
+     {
+      "t": 1.4,
+      "log_step": 73,
+      "text_zh": "观察：log step 73 最后一次reset进入episode 11、resets_left=0，服务返回control_steps=35、done=true，原因为物体距参考或手腕超过0.25米；run.json评分字段sim_steps=35、task_failure=1。该局没有走完要求的99个控制步，且视频末帧仍未带起飞机。这里不将控制步直接换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–0.60 s，MyoHand单手手指在飞机上方及尾部附近收拢，随后空手抬离；局部手指与机体重叠且没有接触日志，无法确认短暂触碰的具体方式，不能据闭指动作认定已抓握飞机。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "ratio",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "airplane_reference_motion_completed",
+   "t": 0.6,
+   "log_step": 73,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.60 s，MyoHand已收拢手指并开始与桌面上的飞机分离，后续视频 0.84–1.40 s 空手悬在飞机上方，未修复抓取。log step 73 记录评分的最后一局在35个控制步时原生失败终止，尚未完成99步循迹运动。"
+  },
+  "unmet": {
+   "airplane_reference_motion_completed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.6,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.32–0.60 s 手指闭合后向上离开，飞机没有被夹持并带离桌面；视频 0.84–1.40 s 手和飞机之间有清晰空隙，飞机仍留在桌面。推断：飞行所需的有效抓持未建立，不能把靠近或短暂擦碰视为先抓住再滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "观察：log step 73 的评分局脚本在控制步索引7–14递增FDS、FDP、FPL、OP屈指肌命令，索引15以后保持；前28行维持预抓取基部目标，随后执行事先计算的IK轨迹，提交中没有依据抓持观测修正动作。视频 0.32–1.40 s 显示闭指和抬手未能带起飞机。推断：这套肌肉与前臂动作未实现有效抓持，属控制失败；现有证据不足以区分具体接触几何、肌肉参数或跟踪误差。"
+   }
+  ],
+  "gap_zh": "无状态日志或接触对，视频初段的手指与飞机局部重叠，无法确定短暂触碰的方式，contact_actual的how保留未知；run.json评分字段未报告track_pos_err、track_rot_err、track_ok_fraction、track_max，四项数值均留null。无成功示范。agent在log steps 78–79自述飞机已被抓起并携带，但评分视频 0.00–1.40 s 中飞机仍留在桌面，节点和现象以视频为准；不采纳该自述中的抓起高度或可达性判断。"
+ },
+ "summary_zh": "评分的第11局中，MyoHand在飞机附近闭指并抬手，但飞机始终留在桌面，飞机循迹飞行节点未完成，最终为0/1。断点在抓持并带起飞机的操作中：有效抓持未建立，服务在35个控制步时因物体距参考或手腕超过0.25米提前终止。视频与agent自述的“已抓起携带”不符；缺少接触日志和精确循迹指标，接触方式及控制子类型仍不能确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "单节点未达成、断点在抓持并带起飞机的操作中、现象没建立接触、根因停在控制，均与视频及 log step 73 的服务读数一致；控制步未换算为视频时间，时钟合规。补充观察：视频 0.6–1.4 s 右侧视角中飞机在桌面上被推动并转向，但仍在桌面，未被带起，不改变标签结论。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:57:12Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--yyrojfo/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-airplane-fly-i00-standard--yyrojfo/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ebe24fef539d8494fe64e55f7067be428013c9e60347fd3ccdcfb8b29a9bd3b8",
+   "video_sha256": "f8ef8c2ed2468b5ef53c368587dc41f2e2061e795d2413c58fbd044d7232b121",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--ajxkisy.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--ajxkisy.json
new file mode 100644
index 0000000000000000000000000000000000000000..db390442e3440d5a69f74d7eef1ab10402182c4f
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--ajxkisy.json
@@ -0,0 +1,199 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-privileged--ajxkisy",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 64,
+   "agent_resets": 412,
+   "agent_sim_steps": 14020,
+   "agent_records": 2,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.942,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.245,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-privileged--ajxkisy/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "135d119ba32acd181ecbbc944acebab4575f0fa720439cd981aca8a60a560ff7"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.92,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–2.28 s，两幅视角中MyoHand手指在红苹果周围屈伸，右侧近景的苹果始终贴近桌面；视频 0.80–1.16 s 手部上移而苹果没有随手升离桌面，随后手指与苹果上缘出现间隙。成功参考视频 0.80–1.16 s 的同一近景能清楚看见苹果离桌，故本次缺少所要求的抬升运动。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 108,
+      "text_zh": "观察：log step 108 对已保存交付文件运行评分器两次，均执行57步且末态哈希相同，报告 on track at 47 of 57 steps, 0.068 m off at worst；log step 112 再次回放结果一致，log step 113 重新生成记录为 track_ok=47、track_max=0.0675、success=False。视频 1.32–2.28 s 苹果仍贴近桌面，末段手指回拢也没有补做完整抬升轨迹。"
+     },
+     {
+      "t": null,
+      "log_step": 108,
+      "text_zh": "推断：交付回放只有47个合格控制步，少于52步，最大误差约0.068米也超过0.05米，因此完整轨迹节点始终不成立；初始位置接近及末段回到附近均不构成该节点达成。track_ok和track_max取上述评分器回放报告，末步track_pos_err没有单独报告。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–1.04 s，MyoHand弯曲的手指贴压、擦过红苹果上部和侧面，苹果在桌面附近轻微移动、转动；视频 1.12–1.40 s 手指抬离，末段又回到苹果周围。可见接触没有形成带动苹果离桌的稳定对向夹持，按实际贴压标注。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok",
+   "value": 47,
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": 0.068,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.84,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.76–1.16 s，MyoHand在苹果周围执行屈指和抬手动作，但红苹果未随手离桌，视频 1.12–1.40 s 手指进一步与苹果分离。log step 108 对交付文件的两次评分器回放均报告47/57步合格、最大误差约0.068米；本步的抬升段没有完成，末段也未修复整段轨迹。视频 0.84 s 为可见偏差的粗略定位，不是精确误差超限时刻。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.92,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.80–1.16 s，右侧近景的红苹果仍在桌面附近，手指上移时没有把它带到抬升位置；视频 1.16–1.40 s 可见苹果与上方手指分离。成功参考视频 0.80–1.16 s 在相同视角显示了明显的苹果离桌运动，本次没有出现该运动。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 80,
+    "t": null,
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent的控制策略未能把手腕和手指运动转化为苹果的承重抬升。log step 72–73 的同参数诊断策略显示手腕上升而苹果高度几乎不变，log step 80 的交付脚本继续使用相同的参考姿态PD肌肉驱动、拇指偏置和腕部反馈；log step 108、112 的交付回放仍失败，且视频 0.80–1.40 s 直接显示手部与苹果运动脱耦。log step 117 中agent也明确自述保留的是苹果不抬升的策略；会话中其他策略的抬升不属于交付轨迹。"
+   }
+  ],
+  "gap_zh": "没有逐步模拟状态及交付回放的接触力记录，无法确定首次厘米级误差超限时刻，也无法将抓持承载不足、接触几何和腕指跟踪偏差可靠区分，故根因只标“控制”。末步track_pos_err未单独报告；47/57和约0.068米来自log step 108、112的评分器回放输出。视频时刻直接取帧时间，不把57个20毫秒控制步或agent工作时间直接换算为本视频时钟。"
+ },
+ "summary_zh": "本次57步交付轨迹中，MyoHand接触并轻微拨动苹果，但苹果始终贴近桌面，未完成“苹果沿轨迹移动”，最终0/1节点。视频约0.84 s的抬升段已经出现手部运动而苹果不随之抬起；交付评分器两次回放均仅47/57步合格、最大误差约0.068米，支持控制失败，后续会话实验的抬升未写入交付策略。缺少逐步状态和接触力，不能进一步确定抓力、接触几何或跟踪误差各自的贡献。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点（0.84 s 抬升段）、位姿不准现象均与视频和评分器回放（47/57、0.068 m）一致；成功参考同视角苹果明显离桌，本次苹果始终贴近桌面。根因只标“控制”偏保守但有 gap 说明，不属于错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:50:35Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--ajxkisy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--ajxkisy/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f66da17f6adf8c719239fc65d8a2ef65858eb8b10dfc391306e40652923d20e3",
+   "video_sha256": "135d119ba32acd181ecbbc944acebab4575f0fa720439cd981aca8a60a560ff7",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--aucxmgh.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--aucxmgh.json
new file mode 100644
index 0000000000000000000000000000000000000000..70278d38b9ceadfff9c2498552afb7b86d31a59c
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--aucxmgh.json
@@ -0,0 +1,227 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-privileged--aucxmgh",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 102,
+   "agent_resets": 81098,
+   "agent_sim_steps": 744160,
+   "agent_records": 0,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.32,
+   "peak_processes": 16,
+   "peak_disk_gb": 0.993,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-privileged--aucxmgh/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "8ed356e15ecf8565b4afa18cd53cc461b375ac20723469aafdf4c71833eda6dd"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.32,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–2.28 s 的两个视角；视频 0.00–0.36 s，MyoHand指端接触苹果上部，苹果只在桌面附近轻微移动。视频 0.64–1.32 s，手指抬起并在苹果上方收拢，苹果仍留在桌面；视频 1.36–2.28 s 也未出现持苹果完成抬升后回落的过程。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：成功参考 inputs/oracle/replay.mp4 的视频 0.64–1.28 s 中，手指围住苹果并使其明显升离桌面后回落；本次视频同段手与苹果分离，未见这一抬升。参考仅用于辨识可见动作，不作为本次达标证据。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 257,
+      "text_zh": "观察：log step 243 生成并保存最终57×45动作数组，其直接回放返回 track_ok=47、末步 track_pos_err=0.00952 m、原生 info.done=False；log step 247、257 对交付文件的官方双次新进程回放均返回 success=false、47/57步在轨、最大误差约0.068 m，且 instance_ok=true、deterministic=true。run.json 的 reward 另记录 n_actions=57、sim_steps=57、replay_success=0。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 257,
+      "text_zh": "推断：交付文件的确定性回放结果与视频中苹果未升起一致；47步少于所需52步，最大误差约0.068 m超过0.05 m，因此整段轨迹验收从未成立。起始对齐或末步误差0.00952 m达标均不能单独构成完成；native_metrics采用上述交付回放结果，最大误差只有评分原因中的舍入值。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.36 s，MyoHand的指端搭压苹果上部，苹果贴近桌面轻微移动；随后手指向上收拢并离开苹果。未见合手夹住苹果并承载抬升，因此实际接触记为指端压触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": 0.00952,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok",
+   "value": 47,
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": 0.068,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.8,
+   "log_step": 257,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约0.64–1.20 s，MyoHand抬起并收拢手指，苹果没有随手升起，而是留在桌面；至视频2.28 s仍未补做该抬升。log step 257对最终交付文件的两次官方回放均确认整段只有47/57步在轨、最大误差约0.068 m。视频0.80 s为可见手物分离的粗略导航点，非精确误差首越界时刻。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": 257,
+    "evidence_zh": "观察：逐帧检查视频0.00–2.28 s的两个视角；视频0.00–0.36 s，MyoHand指端仅搭压苹果上部；视频0.64–1.32 s，手指在苹果上方收拢并抬起，与苹果分离，苹果仍留在桌面。全段未见合手夹住苹果并承载抬升，视频1.36–2.28 s也未补成这一抓持。这里缺少的是带动苹果所需的承载接触；起初的指端压触仍记入contact_actual。log step 257的交付回放报告最大位移误差约0.068 m，仅辅助说明轨迹未达标。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.8,
+    "log_step": 242,
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频0.00–0.36 s指端仅搭压苹果上部，视频0.64–1.20 s手指在苹果上方收拢、与苹果分开，说明执行的指位与腕部运动未形成可承载苹果的围握，腕部抬升因而没有带动苹果。log step 242的最终生成脚本按指关节闭合与腕部运动时序记录动作，log step 243保存后即回放失败；这些代码和交付记录支持这是本次接触控制未奏效。可在同一冻结起点复现的替代指法未核验。"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志，无法精确定位首次厘米级误差越界时刻或核实每个指节的接触力；实际接触方式依据两个视频视角判断。log step 236的接触力检查属于会话诊断，不作为交付轨迹的状态日志；不能仅凭agent在log step 266对关节限位的自述认定任务不可完成。最大误差只有交付回放原因中的约数，具体肌肉参数及可行替代控制未验证。"
+ },
+ "summary_zh": "交付轨迹执行了57步，MyoHand起初以指端压触苹果，随后在苹果上方收拢并抬手，苹果未随手升离桌面，唯一节点“苹果沿轨迹移动”未完成。断点位于该步操作中，现象为没建立接触，接触控制未形成承载抓持；最终交付的官方确定性回放仅47/57步在轨、最大误差约0.068 m。缺少逐步状态日志，首次精确越界时刻、接触力及替代控制的可行性未核实。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点未达、断点、时钟均正确；但现象应为没建立接触而非位姿不准：视频中手指从未围住苹果，且位姿不准只复述了最大误差指标。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "节点未达、断点、现象（没建立接触）与根因（抓取和接触设计，alt_feasible 未知）均有视频与回放记录支持；第一轮的修订已落实，未发现新错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:11:05Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--aucxmgh/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--aucxmgh/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--aucxmgh/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--aucxmgh/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "daab94863b7bebea8ca6f30265b6c41cbec5f1eb67f66d799e7ccb4f920eb733",
+   "video_sha256": "8ed356e15ecf8565b4afa18cd53cc461b375ac20723469aafdf4c71833eda6dd",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--c5okrg4.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--c5okrg4.json
new file mode 100644
index 0000000000000000000000000000000000000000..6be3beb2dfa2476b62b691942d34e534447273d1
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--c5okrg4.json
@@ -0,0 +1,176 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-privileged--c5okrg4",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 0,
+   "agent_resets": 0,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.522,
+   "peak_processes": 20,
+   "peak_disk_gb": 0.876,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-privileged--c5okrg4/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "c06f259ced509ec1f755dfc8fe9c8644be6378870544074b8a12cbd05f0096b4"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [
+     [
+      2.28,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–2.28 s；视频 0.80–1.12 s 的右侧近景显示红苹果随MyoHand升离桌面，拇指与其余弯曲手指从相对两侧围住苹果，苹果随腕部移动；左侧视角显示同一只手和苹果，没有第二只手参与。"
+     },
+     {
+      "t": 2.28,
+      "log_step": null,
+      "text_zh": "观察：视频 1.20–2.28 s 逐帧显示MyoHand带着苹果回落到桌面附近，视频 2.28 s 末帧手指仍围住苹果；未见苹果脱手飞离或掉落。"
+     },
+     {
+      "t": null,
+      "log_step": 125,
+      "text_zh": "观察：log step 125 打印交付文件的生成诊断及两次独立进程回放结果；生成诊断为 steps=57、on_track=57、max_error_m=0.02227、final_error_m=0.00275，两次回放均 success=true、replayed_rows=57、state_sha=758cf104a504519a。上述生成诊断是会话辅助证据，不是逐帧状态日志。"
+     },
+     {
+      "t": 2.28,
+      "log_step": null,
+      "text_zh": "推断：inputs/run.json 的评分器字段 success=1、replay_success=1、deterministic=1、n_actions=57、sim_steps=57、task_failure=0，与视频完整抬升后回落一致，支持唯一目标的整段轨迹验收通过。该判据要求完成全部57步，故到视频 2.28 s 末帧才记成立，不能从视频 0.00 s 的初始对齐记为完成；厘米级误差和覆盖率依赖验收结果，未由画面独立量得。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.80–1.12 s 近景中，单侧MyoHand的拇指和其余弯曲手指从相对两侧围握红苹果，带动其离开桌面；随后随腕部回落，视频 2.28 s 末帧仍围握苹果。两幅图是同一只手的不同视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "MyoHand单手围握苹果，完成抬升和回落；唯一节点“苹果沿轨迹移动”在视频 2.28 s 完整轨迹结束时达成，最终1/1，评分器确认57步交付成功且确定性回放成功。没有状态日志，视频不能独立核验厘米级误差及逐步覆盖率；log step 125 的交付诊断提供辅助数值，但评分器reward字段未报告三个原生指标，故对应value保留null。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "单节点在视频 2.28 s 末帧达成、抓握接触、final=1 与评分器字段一致，未发现错误；validator 通过。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:45:03Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--c5okrg4/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--c5okrg4/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "75e0e2ce584a3911d830a9675faf7b0121238b1763db61ad2cf411c6abf95704",
+   "video_sha256": "c06f259ced509ec1f755dfc8fe9c8644be6378870544074b8a12cbd05f0096b4",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--gckxjfs.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--gckxjfs.json
new file mode 100644
index 0000000000000000000000000000000000000000..77955dbd6ff0cd83e1d81fe86b4c258d955afca4
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--gckxjfs.json
@@ -0,0 +1,203 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-privileged--gckxjfs",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 261,
+   "agent_resets": 2101,
+   "agent_sim_steps": 81526,
+   "agent_records": 2,
+   "agent_replays": 1199,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.254,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.267,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-privileged--gckxjfs/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "171be57e72320c2ee5973ca83cd5ab3930e3e1161925c9a50279eb40ac25b2b6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.2,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.40 s，MyoHand手指在红苹果上部、侧面短暂贴触，苹果仍由桌面承托；视频 0.60–1.20 s手指与苹果逐渐分离，未见合手围握或随手抬升。"
+     },
+     {
+      "t": 0.92,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.76–1.16 s，红苹果留在桌面高度，手悬在其上方；成功参考视频约 0.80–1.12 s则可见红苹果随手离桌并回落，说明本次缺少所需的抬升段。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 371,
+      "text_zh": "观察：视频 1.20–2.28 s直至末帧，红苹果仍在桌面，未补做抬升；log step 370重新生成并保存最终57步动作，log step 371对该交付文件两次独立回放，均为47/57步达标、最大误差约0.068米、success=false，且instance_ok与deterministic均为true。"
+     },
+     {
+      "t": null,
+      "log_step": 362,
+      "text_zh": "观察：log step 362保存的动作回放报告末步track_pos_err=0.01579米、track_ok=47、原生info.done=false，state_sha=51005d4fcda90a95；log step 371最终文件回放得到相同state_sha及失败原因。推断：末步接近参考不能弥补覆盖率不足和最大误差超限，完整轨迹节点从未成立。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.40 s，单侧MyoHand手指在红苹果上部及侧面短暂压触，随后手指离开；两视角未见拇指与其余手指对向夹住苹果，故不记抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": 0.01579
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better",
+   "value": 47
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": 0.068
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.8,
+   "log_step": 371,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80–1.12 s，MyoHand手部在苹果上方移动，红苹果没有升离桌面完成参考抬升；视频至 2.28 s未修复这一缺失。log step 371最终交付文件的回放明确报告47/57步达标、最大误差约0.068米，整段参考跟踪失败。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.92,
+    "log_step": 371,
+    "evidence_zh": "观察：视频 0.80、0.92、1.12 s，红苹果始终停在桌面高度，手指悬于上方，缺少成功参考视频同段可见的抬升位移；log step 371确认最大位置误差约0.068米，超过0.05米上限，且只有47步在0.03米误差内。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 363,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 363的最终生成器以rest_action及当前腕部位置构造动作，只在前10个控制步增加两项无名指屈肌命令，并在控制步15–22增加腕部ARRz旋转；log step 370执行该生成器，log step 371确认文件稳定复现失败。结合视频 0.20 s的短暂压触及视频 0.80–1.12 s手指离开、苹果不抬升，可判定所交付控制未形成带动苹果完成抬升的动作；证据不足以确定更细的控制子类。"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态日志或最终交付动作的接触对、接触力记录；视频可判定缺少抬升，精确误差和覆盖率采用log step 362及371的交付回放反馈，最大误差仅报告到约0.068米。接触设计、腕部轨迹和反馈控制各自的贡献无法进一步分开，断点视频约0.8秒仅作导航锚点，未将会话测试的控制步换算为视频时间。"
+ },
+ "summary_zh": "本次交付执行了57个控制步，MyoHand前段短暂压触苹果，随后手指离开，苹果未完成参考抬升，唯一节点未达成。最终文件两次确定性回放均仅47/57步达标、最大误差约0.068米，失败断点为苹果沿轨迹移动，现象为位姿不准，根因保守归为控制。无逐步状态和接触力记录，无法细分控制原因；成功参考仅用于辨认应有的离桌动作。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点未达成、断点、位姿不准现象和保守的“控制”根因与视频及 log step 362/371 的回放结果一致；未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:47:50Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--gckxjfs/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--gckxjfs/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9761a921ca9c95362b2727fbee1bd1011bd57d9f34284d797d9f57be7b991ebb",
+   "video_sha256": "171be57e72320c2ee5973ca83cd5ab3930e3e1161925c9a50279eb40ac25b2b6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--snwckls.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--snwckls.json
new file mode 100644
index 0000000000000000000000000000000000000000..877e92aa9390a32694ed32722405997a2f3c0e18
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-privileged--snwckls.json
@@ -0,0 +1,196 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-privileged--snwckls",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 463,
+   "agent_resets": 7208,
+   "agent_sim_steps": 283433,
+   "agent_records": 0,
+   "agent_replays": 5,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 9.552,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.842,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 1,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_oom": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-privileged--snwckls/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "d1728a13e8c8c11ff44d0efa64fcb580c5ab2795c680070dea07f657763eb1b0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：逐帧查看视频 0.00–2.28 s；MyoHand初段在苹果上侧活动，随后抬起手指，视频 0.60–1.16 s手指与苹果明显分离，红苹果仍在桌面附近，并未随手升起；视频 1.80–2.28 s手指再次靠近苹果。成功参考视频 0.60–1.16 s则可见苹果随围拢的手指升离桌面，此处仅用于辨识抬升外观。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 208,
+      "text_zh": "观察：log step 203将选定的57步动作保存为最终交付文件，log step 208重新生成该文件并作两次独立评分回放，两次均返回success=false、steps=57、同一state_sha，以及“on track at 48 of 57 steps, 0.069 m off at worst”；inputs/run.json的reward也给出replay_success=0、deterministic=1、instance_ok=1。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 208,
+      "text_zh": "推断：最终文件只有48步满足0.03米带宽，少于要求的52步，最大误差约0.069米也超过0.05米；结合视频中未出现参考抬升，整段跟踪节点从未完成。初始对齐和末段靠近均不能满足整段过程验收。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–0.32 s，单侧MyoHand的指端从红苹果上侧接触、拨动苹果，苹果在桌面附近发生横向移动；视频 0.60–1.16 s手指离开苹果，末段视频 1.80–2.28 s又靠近并拨动苹果。未见手指形成承重夹持并把苹果带离桌面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok",
+   "value": 48,
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": 0.069,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.6,
+   "log_step": 208,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.60 s，MyoHand抬起的手指已与红苹果分开，苹果仍留在桌面附近；视频 0.60–1.16 s没有出现随手抬升，至视频 2.28 s也未补做这一段。log step 208对最终交付文件的两次评分回放均以48/57步合格、最大偏差0.069米判失败；断点取未完成的苹果沿轨迹移动步骤，视频时间仅为可见偏差的粗略锚点。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.6,
+    "log_step": 208,
+    "evidence_zh": "观察：视频 0.60–1.16 s，MyoHand的手指悬在苹果上方，红苹果持续留在桌面附近，没有形成参考抬升所需的向上运动；视频 1.20–2.28 s主要是桌面附近的横向运动和手指重新靠近。log step 208最终文件的评分回放确认过程偏差超限，而非仅由末帧位置判负。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：视频 0.00–1.16 s显示MyoHand的指端虽然作用于苹果，随后却独自抬离，苹果没有随手运动；log step 149的搜索结果及log step 203、208对选定动作的保存和独立回放均显示相同的48/57步、约0.069米峰值偏差，说明交付动作未实现把手部运动传递给苹果的控制。现有证据不能进一步区分接触几何、肌肉激活与腕部跟踪的具体贡献，故只归到控制。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "无状态日志和接触力记录，无法确定逐指接触对、失去接触的精确时刻，或将控制根因细分；断点用视频约 0.60 s的分离现象定位，未将agent时间或会话测试步数换算成视频时间。native_metrics中的48步和0.069米取自log step 208对最终文件的评分回放reason；该回放未单独报告末步track_pos_err，故其value为null，log step 203的末步误差为候选文件自存统计，不作为该项评分器读数。"
+ },
+ "summary_zh": "观察：本次交付完整回放57个控制步，但视频中MyoHand抬离时红苹果仍留在桌面附近，唯一的整段轨迹节点未完成，最终为0/1。推断：断点在苹果沿轨迹移动的操作中，现象为位姿不准，根因保留为控制；log step 208两次评分回放均为48/57步合格、最大误差约0.069米。观察：缺少状态和接触力日志，具体控制失效机制及最终回放的末步误差读数不能独立核验。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "终局0/1、节点未完成、断点在唯一节点、log step 208的48/57步与0.069米读数均与证据一致；视频与评分误差曲线（第19–28步超限，约视频0.8–1.1 s）吻合断点锚点。现象位姿不准、根因控制有证据支持，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:49:49Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--snwckls/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-privileged--snwckls/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "54c359c49e619803938bfaa013bed3312a99a31f3f8718d0837b60ab88201db2",
+   "video_sha256": "d1728a13e8c8c11ff44d0efa64fcb580c5ab2795c680070dea07f657763eb1b0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--62kiiix.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--62kiiix.json
new file mode 100644
index 0000000000000000000000000000000000000000..e39f7ba55c5f69121aaab79901ec49d41d959716
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--62kiiix.json
@@ -0,0 +1,215 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--62kiiix",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "physics_steps": 570,
+   "steps": 477,
+   "calls": 225,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 477,
+   "elapsed_s": 1013.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.335,
+   "peak_processes": 8,
+   "peak_disk_gb": 0.247,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--62kiiix/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "06c69324ba3cf67d9efee6ce72336b7d80d1689fa0c12a914437d4a5b89a0fd9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.48 s，MyoHand手指在苹果上方及侧面收拢，苹果在桌面上向手外侧移动；视频 0.24 s 后近景可见手指与苹果之间的间隙，未见对向夹紧后带起苹果。"
+     },
+     {
+      "t": 0.92,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频 0.00–2.28 s；视频 0.80–1.16 s 苹果仍贴近桌面，手在其侧上方空握，未出现抬升；视频 1.32–2.28 s 也未重新抓起。成功参考视频 0.80–1.16 s 则可见苹果随手升起、底部与桌面分离，随后回落；参考仅用于识别可见线索。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 51,
+      "text_zh": "观察：log step 51 重置后运行最后一次控制器，服务终结原因为“on track at 47 of 57 steps, 0.068 m off at worst”，并报告 episode=11、control_steps=57、resets_left=0；log step 53 再次确认相同终局。评分字段 graded_by_live=1、n_actions=57、task_failure=1，故本视频对应最后一局，先前重置掉的测试不计进度。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 51,
+      "text_zh": "推断：本节点要求完整轨迹验收，初始误差为零及末尾接近原位均不单独成立；最后一局服务报告的47步少于52步，最大误差0.068米大于0.05米，且视频未见抬升，故全程没有达成本节点。时间锚点直接取视频帧，未把agent用时或20毫秒控制步直接换成视频秒。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–0.20 s，单侧MyoHand的指端贴近并拨动红苹果上部及侧面，苹果沿桌面向手外侧移动；随后手指空收拢并离开苹果。未见把苹果对向夹紧的有效抓握，实际接触按指端拨推标注。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok",
+   "value": 47,
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": 0.068,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.24,
+   "log_step": 51,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.16–0.32 s，MyoHand收指时苹果仍在桌面，手与苹果分离；后续视频 0.80–1.16 s 的抬升阶段手未带起苹果，直到视频 2.28 s 仍未修复。log step 51 的最后一局服务判为47/57步达标、最大误差0.068米；断点取抓握未建立、手物开始明显分离处。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.24,
+    "log_step": 51,
+    "evidence_zh": "观察：视频 0.00–0.20 s 手指只拨动桌面上的苹果，视频 0.24–0.80 s 手在苹果外侧空收拢；视频 0.80–1.16 s 及末段视频 1.32–2.28 s 均未见苹果被手夹住、随手抬离桌面。推断：用于承载苹果的有效抓握从未建立，早期指端触碰不等于握住后再滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 51,
+    "source_zh": "日志",
+    "evidence_zh": "观察：log step 44 的控制器以手指关节速度低于STALL_V作为contact标志，log step 48 又加入全部stall后退出夹紧阶段的条件；log step 51 最后一局打印finger contact及squeeze complete，继续执行腕部抬升。视频 0.24–1.16 s 显示手指已在苹果外空收拢，苹果没有随手离桌。推断：控制器把关节停滞当作成功接触，未确认承载抓握便进入抬升，未能协调闭指与腕部运动；证据支持控制问题，尚不足以进一步确定接触几何或单个肌肉参数的责任。"
+   }
+  ],
+  "gap_zh": "没有逐控制步状态、接触对或接触力记录，不能确定每根手指的接触时刻和受力，也不能独立恢复全部误差曲线。track_ok=47与track_max=0.068取自log step 51及53中最后一局服务的终结结果，后者为舍入值；评分记录未报告末步track_pos_err，故留null，未采用agent图像估计或其最终自述的0.9厘米。根因仅确定到控制层级。"
+ },
+ "summary_zh": "最后评分局完整运行57个控制步，但苹果仅被指端拨动，未被有效抓握并抬离桌面，苹果沿轨迹移动节点始终未达成，最终0/1。视频约0.24 s手物分离后未修复；最后一局服务报告47/57步达标、最大误差0.068米，结合脚本以关节停滞判断接触的逻辑，归因为控制未能建立并确认抓握。没有逐步状态及接触力记录，无法细分肌肉参数责任或给出末步原生位置误差。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、final=0、断点在唯一步骤、最后一局由服务字段（47/57步、最大0.068米）界定，均有视频和日志支持；未发现需要改内容的错误。现象与根因（没建立接触/控制）证据足够，未改动。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:51:32Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--62kiiix/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--62kiiix/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3b973e055219c935ee49be46720fd0e488d4e36e93b895e4317a3bd7adee8524",
+   "video_sha256": "06c69324ba3cf67d9efee6ce72336b7d80d1689fa0c12a914437d4a5b89a0fd9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--agetblx.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--agetblx.json
new file mode 100644
index 0000000000000000000000000000000000000000..64adc0661cadb847b7d79282fbc2fdfff4c9e730
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--agetblx.json
@@ -0,0 +1,216 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--agetblx",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "physics_steps": 570,
+   "steps": 114,
+   "calls": 4,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 114,
+   "elapsed_s": 3499.6,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.124,
+   "peak_processes": 20,
+   "peak_disk_gb": 0.375,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--agetblx/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "2ff11b099e4dce14045c20065e7ac11fd2b5af592cc8ae6d46b55fd7d8365967"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.48,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.44 s，MyoHand指端贴近并推拨红苹果上侧，苹果沿桌面移动；视频 0.48–0.64 s，手指已在苹果上方收成空拳，苹果没有随手升起。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：视频 0.80–1.52 s，红苹果继续沿桌面滚离拳头；视频 1.32–2.28 s逐帧检查仍未见苹果被重新握起。成功参考视频 0.80–1.16 s的同一近景能清楚看到苹果随手离桌升起，本次未出现这一运动。"
+     },
+     {
+      "t": null,
+      "log_step": 67,
+      "text_zh": "观察：log step 67，agent脚本重置后向服务提交57行执行器命令；服务末回合返回control_steps=57、resets_left=9，原因是“on track at 18 of 57 steps, off track at the end, 0.092 m off at worst”。这是评分回合；log step 54的12/57步结果属于已被重置的前一回合。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 67,
+      "text_zh": "推断：视频 2.28 s仍无完整抬升回落动作，且log step 67的末回合服务判据明确不满足至少52步达标、末步达标及最大误差不超过0.05米，因此整段轨迹节点从未成立；不能将开局的瞬时接近或18个达标采样点当作完成节点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–0.44 s，MyoHand指端在红苹果上侧接触并推拨，苹果沿桌面移动；随后手指在苹果上方收成空拳，未见拇指与其余手指合围夹持苹果并带其抬升。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better",
+   "value": 18
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": 0.092
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.48,
+   "log_step": 67,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约0.48 s，MyoHand已在红苹果上方收成空拳，苹果仍在桌面，随后滚离手部；至视频 2.28 s也未被重新握起。log step 67确认该末回合执行完57个控制步但轨迹验收失败。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.48,
+    "log_step": null,
+    "evidence_zh": "推断：视频 0.00–0.64 s虽有MyoHand指端对苹果的推拨接触，但没有形成可承载苹果的合围抓握；手指收成空拳时苹果留在桌面。视频 0.80–2.28 s苹果沿桌面滚动、手部与苹果分离，未见一次持物抬升，故所需抓握未建立。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 0.48,
+    "log_step": 67,
+    "source_zh": "视频、日志",
+    "evidence_zh": "推断：视频 0.32–0.64 s，MyoHand屈指和腕部运动未使苹果进入闭合手指之间，反而让苹果沿桌面移出；log step 67的脚本把腕部与肌肉命令作为一个57步序列提交，服务返回18/57步达标、末步离轨。控制序列未能建立并带动持物运动，足以支持控制层面的归因；没有可靠接触力或逐步关节/物体状态，不能进一步确定为握法几何、控制跟踪或某个参数问题。"
+   }
+  ],
+  "gap_zh": "观察：本目录inputs/提供本次及成功参考视频和run.json，但没有状态日志；log step 67的服务终结原因只报告达标步数、最差误差及末步未达标，没有末步track_pos_err数值，故该指标填null。视频能判定未握起苹果，但不能精确测量每步厘米级误差、接触力或区分具体控制子类；track_ok=18与track_max=0.092取自该末回合服务终结原因，后者仅有三位小数精度。"
+ },
+ "summary_zh": "评分回合中，MyoHand指端推拨苹果后收成空拳，苹果沿桌面滚离，未完成随参考抬升再回落的轨迹，节点最终为0/1。log step 67的服务结果为57步中18步达标、末步离轨、最大误差0.092米；断点是苹果沿轨迹移动中的抓握未建立，根因可支持到控制层级。缺少逐步状态和接触力记录，无法进一步确定控制子类或给出末步精确位置误差。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "评分回合（log step 67，服务返回18/57步达标、末步离轨、最差0.092 m）与视频一致：苹果被推离、手收成空拳，节点未达成；断点、现象与控制归因均有证据支持，未见需要修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:04Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--agetblx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--agetblx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d230178225950af61dc52152b2273efe2184dbb44f6f4ac21398bd6164ed5e3c",
+   "video_sha256": "2ff11b099e4dce14045c20065e7ac11fd2b5af592cc8ae6d46b55fd7d8365967",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--crrjuej.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--crrjuej.json
new file mode 100644
index 0000000000000000000000000000000000000000..16cd441102072a8530504f225bdbb47e459c7688
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--crrjuej.json
@@ -0,0 +1,208 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--crrjuej",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "physics_steps": 570,
+   "steps": 166,
+   "calls": 14,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 166,
+   "elapsed_s": 2682.9,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.52,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.292,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--crrjuej/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "bca38c5fe00a896db2aaf6b40fa7e7b3118caf7781bfaf954c8a114bf25f22b5"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–2.28 s；视频 0.40–0.64 s，MyoHand手指在苹果上侧闭合，苹果沿桌面向手外偏移；视频 0.76–0.92 s，手指与苹果逐渐分离，未见围握带动苹果升离桌面的过程。"
+     },
+     {
+      "t": 2.28,
+      "log_step": null,
+      "text_zh": "观察：视频 0.96–2.28 s，红苹果在桌面上滚向手外并停留，手在另一侧空握，直至末帧也未重新抓回；成功参考视频 0.80–1.24 s可见苹果随手升起再回落。本次评分记录为n_actions=57、live_success=0、replay_success=0、live_matches_replay=1，完整参考轨迹未完成。"
+     },
+     {
+      "t": null,
+      "log_step": 92,
+      "text_zh": "观察：log step 92的agent脚本重置后向服务提交57行肌肉及腕部控制；log step 93的服务输出确认episode=3、resets_left=8、control_steps=0的起点；评分记录共3局、2次重置，最后一局为评分局。此前两局的调试动作不计入本次进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.40–0.76 s，MyoHand指端从苹果上侧接触并将其拨向手外；随后苹果滚离。未见拇指与其余手指形成夹持苹果并随手搬运的围握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.64,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.48–0.88 s，MyoHand闭指时苹果沿桌面移出手指包围区，手随后抬动而苹果滚离；视频 0.96–2.28 s始终没有重新抓取并恢复参考运动。失败发生在苹果沿轨迹移动这一步的抓持尝试中。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.64,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.32–0.88 s，MyoHand手指虽然碰到苹果上侧，却没有形成带动苹果抬升所需的对向围握；苹果从指端下方向外滚出。视频 0.96–2.28 s，手部空握，苹果留在桌面远处，未见先抓稳再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 92的agent脚本已按参考位移生成腕部运动，并以预设肌肉激活斜坡闭指，一次提交57行控制；视频 0.40–0.92 s却显示闭指将苹果拨出，未形成承载抓持。目标是抓持后跟随参考，但肌肉与腕部协调未实现所需接触，支持归因于控制；现有证据不足以进一步区分接触设计、动力学建模或反馈控制子类。"
+   }
+  ],
+  "gap_zh": "没有模拟状态日志或接触对记录，无法精确核验逐控制步接触力及厘米级轨迹误差；评分字段未报告track_pos_err、track_ok、track_max，故三项原生指标均填null。log step 105为agent自己的视觉与深度拟合估计，不当作评分器测量；控制问题的具体子类尚不能确定。"
+ },
+ "summary_zh": "本次评分局完成了57个控制步，但MyoHand闭指时把苹果拨向手外，随后苹果沿桌面滚离，完整参考轨迹节点未达成（终态0/1）。断点约在视频 0.64 s，所需围握没有建立，日志与视频支持控制层面的失败。没有状态接触记录和原生误差指标，无法确定具体控制子类或精确超限时刻。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频一致：0.4–0.9 s 闭指把苹果拨向手外，随后苹果滚离，绿色参考球上升而苹果未跟随；节点未达成、断点、没建立接触、控制（子类不明）均有证据支持。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:49:50Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--crrjuej/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--crrjuej/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "34372de9c12b7621cc8574b47c4ecb0c944f804648a52ba1eeded8a292391f33",
+   "video_sha256": "bca38c5fe00a896db2aaf6b40fa7e7b3118caf7781bfaf954c8a114bf25f22b5",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--eyv5pym.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--eyv5pym.json
new file mode 100644
index 0000000000000000000000000000000000000000..fedc1dbfe24da8190179c875d5c0c3c8fbcaf7ea
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--eyv5pym.json
@@ -0,0 +1,190 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--eyv5pym",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "physics_steps": 570,
+   "steps": 538,
+   "calls": 1088,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 538,
+   "elapsed_s": 3328.4,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.033,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.699,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--eyv5pym/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "eccc28ddaeb686ad61d2d9b3c8f6ae9ddacd4569bf25821b6a899491f111cf72"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [
+     [
+      2.28,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 红苹果位于桌面，MyoHand手指位于其上方和两侧；此时尚未执行完整抬升轨迹，不能凭初始相对位移为零记为完成。"
+     },
+     {
+      "t": 1.0,
+      "log_step": null,
+      "text_zh": "观察：逐帧查看视频 0.00–2.28 s；视频约 0.60–1.00 s，MyoHand弯曲手指围夹苹果并随腕部抬升，视频 1.00 s 右侧近景可见苹果底部与桌面之间的间隙，左侧视角可见苹果随手移动；视频约 1.32 s 苹果回到桌面，随后至视频 2.28 s 留在手指之间的桌面位置。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 142,
+      "text_zh": "观察：视频 2.28 s 为末帧，苹果已回落；log step 131启动最后一次重置后的第11局，log step 134、136、138、140连续报告该局控制步18、27、39、45，log step 142的机器人服务返回control_steps=57、episode=11、done=true、reason=success。inputs/run.json的评分器字段同时记录live_success=1、graded_by_live=1、n_actions=57、sim_steps=57、task_failure=0；之前被重置的调试局不属于本视频的评分尝试。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 142,
+      "text_zh": "推断：视频中抬升后回落的过程与完整57步的服务成功验收一致，故“苹果沿轨迹移动”在视频末帧 2.28 s 完成并保持为真；不把中途抬起或初始对齐单独记为该节点成立。厘米级误差、至少52步达标及最大误差上限依据本次评分器成功判定，画面不能独立精确量距。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 0.60–1.20 s，单侧MyoHand的拇指与弯曲的其余手指从苹果两侧围夹，苹果随手腕升离桌面再回落；是同一只手的围握接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "最后一次重置后的第11局完成苹果抬升、回落及全部57个控制步，服务在log step 142返回success，唯一目标节点在视频末帧 2.28 s 达成，最终为1/1。实际接触为单侧MyoHand手指围握苹果，本次评分尝试成功，无失败断点。没有状态日志，评分器也未报告track_pos_err、track_ok、track_max的具体数值，厘米级轨迹验收依据成功结果，原生指标值保留为null。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "视频 0.6–2.28 s 苹果被单手围握抬起后回落，评分器 live_success=1，唯一目标节点末帧达成、无断点，标签与证据一致。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:47:15Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--eyv5pym/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--eyv5pym/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "2999c10dd7a95dedab9839c8a5c43a8d1fee6b2a701d8b8dea2dc7b8f06a9c24",
+   "video_sha256": "eccc28ddaeb686ad61d2d9b3c8f6ae9ddacd4569bf25821b6a899491f111cf72",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--sat3nwy.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--sat3nwy.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a59bc25f44ccf42a3a03ed3c19b44bd15f95dba
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--sat3nwy.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--sat3nwy",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "physics_steps": 570,
+   "steps": 392,
+   "calls": 174,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 392,
+   "elapsed_s": 1176.2,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.449,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.266,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--sat3nwy/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "317c715703a3374f2e815241f07eba65781f0c6b5980790bd3a021ead955898a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.16 s，MyoHand的手指贴在苹果上方及侧面；视频 0.20–0.28 s，手向后上方移动、手指收拢，右侧近景可见手与苹果分离，苹果仍留在桌面，没有被夹持着带走。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查本次视频全部58帧。视频 0.32–1.28 s，手呈空握状、苹果留在其下方桌面；视频 1.32–2.28 s 末帧期间也未形成持物抬升。成功参考视频 0.72–1.28 s 则可见苹果夹在手指之间并升离桌面，再于参考视频约 1.44 s 回落；该参考仅用于辨认成功动作。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 129,
+      "text_zh": "推断：log step 129 显示服务重置至第11局且重置次数耗尽，inputs/run.json 的评分记录为 episodes=11、n_actions=57、sim_steps=57、live_success=0、task_failure=1；结合视频全程未出现参考所需的持物抬升，完整轨迹节点从未达成，末值为假。初始瞬时对齐不能满足完整57步验收。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.16 s，单侧MyoHand手指贴压苹果上部及侧面，未见拇指与其余手指形成可靠的对向夹持；视频 0.20–0.28 s 起手指在苹果上方空合，未带起苹果。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.24,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.20–0.28 s，MyoHand收拢手指时手已经移离苹果，右侧近景出现明确间隙，苹果仍在桌面；此后直到视频 2.28 s 均未重新握住并带起苹果。失败发生在执行苹果参考轨迹的抓持操作中。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.24,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.16 s 虽有手指贴触苹果，视频 0.20–0.28 s 手指却在苹果上方空合；视频 0.32–2.28 s 苹果没有随手升起。所需的对向抓持接触未建立，未见先持起再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 127,
+    "evidence_zh": "推断：log step 127 的最终脚本预先生成全部57步的腕部位置命令和肌肉激励，并一次提交，执行中没有依据苹果是否被夹住来修正动作；log step 128–129 显示该脚本启动第11局。视频 0.20–0.28 s 可见手指在移离苹果后空合，此后未恢复抓持，支持腕部运动与手指接触控制未协调成功；现有证据不足以进一步区分接触设计、轨迹跟踪或具体参数问题。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "没有状态日志或接触力记录，不能核实每根手指的接触对及失去接触的具体力学机制。评分reward未报告最终第11局的track_pos_err、track_ok、track_max，日志也未给出该局完成后的服务meta，故三个原生指标均为null；log step 129 中47/57步和最大误差0.069 m明确属于重置前第10局，log step 135 将47/57归于第11局只是agent自述，不作为本次指标。视频可判定未建立持物抬升，但不能独立量出逐步厘米级误差。"
+ },
+ "summary_zh": "本次评分为最后的第11局，执行了57个控制步，但苹果沿轨迹移动节点未完成，最终为0/1。视频约 0.24 s，MyoHand手指已移离苹果并空合，后续一直未带起苹果；日志与画面支持抓持控制未协调成功。缺少最终局的精确轨迹误差及接触记录，未将第10局的47/57步结果移用于本局。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "单节点未达成、最终0、断点在 apple_reference_tracked 的判断与视频一致：手在约0.2–0.3 s 离开苹果，苹果全程留在桌面，未见参考所示的升起回落。接触、现象（没建立接触）和根因（控制，停在大类）有证据支持；原生指标为 null 的处理合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:46:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--sat3nwy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--sat3nwy/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "e54f6969ced1359e1519c991abcbba3456831ec840e3501d90080b3e520bc95f",
+   "video_sha256": "317c715703a3374f2e815241f07eba65781f0c6b5980790bd3a021ead955898a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--snevlcm.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--snevlcm.json
new file mode 100644
index 0000000000000000000000000000000000000000..378e7c80510b317f947eb253026c116b546457b4
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--snevlcm.json
@@ -0,0 +1,216 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--snevlcm",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 40,
+   "sim_steps": 40,
+   "physics_steps": 400,
+   "steps": 323,
+   "calls": 166,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 323,
+   "elapsed_s": 1539.5,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.309,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.833,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--snevlcm/replay.mp4",
+  "duration_s": 1.64,
+  "fps": 25.0,
+  "frames": 41,
+  "sha256": "df786ddd53dbde1f76eb5b496ba6463c0dfc0a1bde43ef5e1c5abd53a9067b30"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：视频0.00–0.44 s，MyoHand的指尖贴在红苹果上部，苹果沿桌面移动；未见拇指与其余手指对向夹紧并带起苹果。"
+     },
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：视频0.48–0.80 s，手指逐渐弯曲并离开苹果上方，苹果仍在桌面上滑动；视频0.84–1.60 s苹果持续远离手，右侧近景末段仅能看到苹果的上部，左侧视角仍能看到苹果与手分离。整段未出现参考抬升后回落。"
+     },
+     {
+      "t": 1.6,
+      "log_step": 108,
+      "text_zh": "观察：log step 108启动最终控制器并有一次reset，log step 109继续提交动作，此后没有reset；评分器reward记录本次n_actions=40、sim_steps=40、task_failure=1、deterministic=1、live_matches_replay=1。"
+     },
+     {
+      "t": 1.6,
+      "log_step": null,
+      "text_zh": "推断：评分局仅执行40个控制步，未完成节点要求的57步，且视频显示苹果未随手抬起，因此整段轨迹验收从未成立；不能把初始位置相符计为节点达到。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频0.00–0.44 s，单侧MyoHand指尖压在红苹果上部并随手运动擦推苹果，随后指尖与苹果分离；未见形成对向夹紧的抓握，故不标抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.6,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约0.48–0.80 s，MyoHand弯指并上移时，红苹果仍沿桌面滑走，没有随手升起；到视频1.60 s仍与手分离，没有修复。评分器记录最终仅执行40个控制步，未完成57步参考轨迹。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.6,
+    "log_step": null,
+    "evidence_zh": "观察：视频0.00–0.44 s可见指尖触及苹果上部，但未见对向夹紧；视频0.48–0.80 s手指在苹果上方弯曲，苹果留在桌面并滑离，视频0.84–1.60 s也未被重新抓起。所需的可承载抬升握持未建立，已有的表面触碰不等于抓住。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.6,
+    "log_step": null,
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频0.00–0.44 s实际接触集中在苹果上部，未形成拇指与其他手指的对向夹持；视频0.48–0.80 s手指弯曲和手腕上移时，手与苹果分离而苹果沿桌面滑走。此次接触配置未能承载苹果完成抬升；这是从接触几何和后续运动推断，不能进一步确定具体肌肉参数或反馈增益。"
+   }
+  ],
+  "gap_zh": "没有逐步模拟状态及接触力记录，无法精确确定各手指的接触对和轨迹误差越限的首个控制步；评分字段未报告track_pos_err、track_ok或track_max，三项数值均留null。log step 107仅保留编辑调用片段，未提供最终gen.py的完整代码，log step 108–111也没有服务输出正文；log step 112所称“控制步40越过0.25米终止界限”是agent自述，不能独立核实具体原生终止条件，亦不能据此细分控制参数问题。",
+  "unmet": {
+   "apple_reference_tracked": false
+  }
+ },
+ "summary_zh": "评分局中MyoHand指尖接触并擦推了苹果，但未形成能抬升苹果的握持，苹果沿桌面滑离；“苹果沿轨迹移动”节点始终未达到，终态0/1。断点在视频约0.60 s的轨迹操作中，现象为没建立所需握持，根因推断为抓取和接触设计不适配；评分局40个控制步即结束，未执行完整57步。缺少逐步状态、原生误差指标和最终控制器完整代码，无法核实精确越限时刻或具体肌肉参数。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频0–1.6 s与标签一致：指尖压在苹果上部，苹果被擦推滑离、绿色参考点与苹果分离，节点未达成；断点、现象、根因与证据均有依据，未发现需改动之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:49:35Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--snevlcm/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--snevlcm/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "977ee421f9168ee6af96aed5ff8d2f75e36e8a3d839956596b0923dd4cb35b98",
+   "video_sha256": "df786ddd53dbde1f76eb5b496ba6463c0dfc0a1bde43ef5e1c5abd53a9067b30",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--uar8kf2.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--uar8kf2.json
new file mode 100644
index 0000000000000000000000000000000000000000..936301cee32306545b4fb541852664e4b7f6231f
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--uar8kf2.json
@@ -0,0 +1,238 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--uar8kf2",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "physics_steps": 570,
+   "steps": 532,
+   "calls": 66,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 532,
+   "elapsed_s": 2140.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.611,
+   "peak_processes": 11,
+   "peak_disk_gb": 1.251,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--uar8kf2/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "155ac17773d93eee63421dd0b1f1b622fd7b92f4c2b9f5b914e45ce4c914fbc0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.72,
+      "log_step": null,
+      "text_zh": "观察：视频 0.60–0.72 s，右侧近景可见MyoHand弯曲手指围住红苹果，苹果离开桌面并随手抬起；视频 0.80–0.88 s，苹果向下离开手指，指端与苹果之间出现明显空隙。"
+     },
+     {
+      "t": 2.28,
+      "log_step": null,
+      "text_zh": "观察：视频 1.28–2.28 s，苹果在桌面侧向滚动，MyoHand未重新夹住苹果，末帧苹果仍在手外。已逐帧检查视频全程，尤其是视频 0.76–1.24 s 的分离过程与视频 1.32–2.28 s 的末段。"
+     },
+     {
+      "t": null,
+      "log_step": 241,
+      "text_zh": "观察：log step 241 中，agent脚本重置后执行最后一轮，机器人服务返回 episode=11、control_steps=57、done=True，原生验收原文为“failure: on track at 21 of 57 steps, off track at the end, 0.079 m off at worst”；这是一轮已执行完毕的失败轨迹，不是此前各轮或本地模拟的成绩。"
+     },
+     {
+      "t": null,
+      "log_step": 241,
+      "text_zh": "推断：最后一轮仅21步在容差内，少于所需52步，且末步不达标、最大误差0.079米超过0.05米，因此整段轨迹节点从未成立；初始对齐与短暂带动苹果不满足该节点的完整57步判据。视频锚点直接取所查帧的时间，未将控制步的仿真秒数换作视频秒数。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.60–0.72 s，单侧MyoHand的弯曲手指围住红苹果并将其带离桌面，苹果随手运动；视频 0.80–0.88 s，苹果从指端脱离下落，此后落回桌面并滚开。该围握仅短暂保持。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok",
+   "value": 21,
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": 0.079,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.8,
+   "log_step": 241,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80 s，MyoHand仍在执行手指与腕部运动，苹果已向下离开指端；视频 0.88–1.24 s，苹果落向桌面并横向滚开，此后至视频 2.28 s 未被重新控制。log step 241 的机器人服务终结记录确认这轮57步轨迹仅21步达标，末步偏离参考。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 0.88,
+    "log_step": 241,
+    "evidence_zh": "观察：视频 0.60–0.72 s，红苹果被MyoHand弯曲手指围住并带离桌面；视频 0.80–0.88 s，苹果向下脱离指端并出现明显空隙，随后落回桌面并向侧方滚动；视频 1.28–2.28 s，苹果始终在手外，未被重新握住。log step 241 的服务验收记录报告21/57步达标、末步离轨及最大位移误差0.079米。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 0.8,
+    "log_step": 241,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 241 的最终控制脚本让手指与腕部跟踪参考关节运动，分块发送执行器命令，并用服务返回的本体感知更新本地模拟；视频 0.72–0.96 s 却显示苹果与指端分离，随后滚开且未被重新控制。结合该轮服务记录的21/57步达标，可归因于指腕控制未能稳定带动苹果沿目标轨迹运动；现有证据不足以进一步区分接触几何、肌肉控制跟踪或模型误差。"
+   }
+  ],
+  "gap_zh": "没有状态日志或逐控制步的苹果位姿、接触对与接触力；视频已显示苹果短暂离桌随手抬升后脱落，但无法进一步确定围握失效的具体接触或力学机制，也无法精确定位首次超出厘米级容差的控制步。断点采用视频约0.80秒开始分离的粗略锚点，视频0.88秒可见明确空隙。log step 241 的服务终结记录给出track_ok=21和track_max=0.079米，但未报告末步track_pos_err的具体数值，故该指标留null。仿真57步共1.14秒而视频为2.32秒，未据此臆造逐步时间映射。"
+ },
+ "summary_zh": "最后评分轮中，MyoHand短暂围握并带起苹果，但视频约0.80–0.88秒苹果脱离手指，随后落回桌面向侧方滚动，唯一节点“苹果沿轨迹移动”未完成。log step 241 的机器人服务验收为21/57步达标、末步离轨、最大误差0.079米；结合视频，支持没保持住及控制失败的判断。缺少状态和接触力日志，不能进一步确定接触失效机制或首次超限时刻。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 apple_reference_tracked（视频约0.8 s）与log step 241的21/57服务读数均成立；但视频0.0–0.8 s苹果被手指带离桌面随手上升，随后才掉落，应为没保持住而非位姿不准，且接触应为手指带起而非推/扫。",
+    "n_issues": 2,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 apple_reference_tracked、没保持住、手指抓握接触及 log step 241 的21/57服务读数均与视频和日志一致，时间均为视频钟，未发现需改动之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:11:42Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--uar8kf2/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--uar8kf2/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--uar8kf2/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--uar8kf2/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "37c2e6a305360da7e1b61240b23337daaa5d0c4e27137f7b1dc06fc34354d6a6",
+   "video_sha256": "155ac17773d93eee63421dd0b1f1b622fd7b92f4c2b9f5b914e45ce4c914fbc0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--yf45bgc.json b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--yf45bgc.json
new file mode 100644
index 0000000000000000000000000000000000000000..ac8fa6e1427aeefdef3624780e7150ae4ce99318
--- /dev/null
+++ b/phase2/myosuite/dex_apple_lift/trajectories/myosuite-dex-apple-lift-i00-standard--yf45bgc.json
@@ -0,0 +1,215 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-apple-lift-i00-standard--yf45bgc",
+ "taskKey": "myosuite/dex_apple_lift",
+ "dag_sha256": "a041469a6976fab4ca5933b2e440cb573fe3bbf63067791e66b2f5edc9daae3e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 57,
+   "sim_steps": 57,
+   "physics_steps": 570,
+   "steps": 134,
+   "calls": 7,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 134,
+   "elapsed_s": 1393.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.837,
+   "peak_processes": 21,
+   "peak_disk_gb": 0.374,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-apple-lift-i00-standard--yf45bgc/replay.mp4",
+  "duration_s": 2.32,
+  "fps": 25.0,
+  "frames": 58,
+  "sha256": "89df15ba93889bbcc7d09796bc8d63d5f5abae1d87120bd387a0d667cbd4341f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "apple_reference_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 70,
+      "text_zh": "观察：log step 70 的服务返回 episode=3、control_steps=57、done=True，判负原因为仅47/57步在轨、最大误差0.070米；log step 80再次查询仍为同一终结状态。本视频只对应该末次episode，不计被reset的前两次尝试和后续离线搜索。"
+     },
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频0.00—2.28 s；视频0.00—0.60 s右侧近景中，MyoHand手指先贴在苹果上部，随后手指与苹果之间出现间隙，红苹果仍贴近桌面；视频0.80—1.20 s手指再次向苹果上方弯曲，苹果没有随手升离桌面。"
+     },
+     {
+      "t": 2.28,
+      "log_step": 70,
+      "text_zh": "推断：结合视频全程未出现承载抬升和log step 70的服务终结判据，完整轨迹节点从未成立，终值为false；47个瞬时容差合格采样及初始对齐不等于完整轨迹完成。服务记录可确定track_ok=47、track_max约0.070米，未给出末步track_pos_err的数值。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：成功参考inputs/oracle/replay.mp4在其视频0.80—1.04 s的右侧近景清楚显示手指围住红苹果、苹果与桌面之间出现离桌间隙，随后回落；本次视频相同阶段没有这一抬升表现。参考仅用于确认可见线索与可行替代，不作为本次进度证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频0.00—0.20 s及0.72—1.04 s，单侧MyoHand的指端贴压、擦过红苹果上部；视频0.36—0.60 s可见手指离开而苹果留在桌面。未看到围合夹住并承载苹果的抓握，故按实际表面接触标注。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok",
+   "value": 47,
+   "unit": "control_steps",
+   "threshold": 52,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": 0.07,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_reference_tracked",
+   "t": 0.6,
+   "log_step": 70,
+   "where": "操作中",
+   "evidence_zh": "观察：视频0.36—0.60 s，MyoHand手指抬动时没有带走桌面上的红苹果；视频0.80—1.20 s仍未抬起，后续直至视频2.28 s也未补成参考抬升。log step 70的末次episode完整执行57步后被服务判为仅47步在轨、最大误差0.070米。"
+  },
+  "unmet": {
+   "apple_reference_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.6,
+    "log_step": null,
+    "evidence_zh": "观察：视频0.00—0.60 s，MyoHand指端先触到苹果上方，随后手部向上运动并与苹果拉开间隙，苹果留在桌面；视频0.80—1.20 s再贴近时仍未带起苹果。推断：该次尝试没有建立抬升所需的承载抓握；这是有表面触碰但未形成有效夹持，不是先成功持物再掉落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "已验证有",
+    "evidence_zh": "推断：本次视频0.00—1.20 s显示指端主要作用于苹果上部，手指抬动时与苹果分离，所形成的接触布局不能带动苹果离桌。log step 46的mk4脚本把拇指、食指和中指的期望指端力指向苹果质心并追加向上分量；log step 69将该参数组合生成plan_fb2，log step 70提交到末次episode后仍无抬升，与视频吻合，支持抓取与承载接触未建立这一原因。成功参考视频0.80—1.04 s确实能围握并抬起同类苹果，证明存在可行替代；没有据此认定本次需要某个未验证的具体参数。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "没有逐帧状态或接触力日志；两视角足以判断苹果未随手抬升，但不能精确恢复各指接触点、法向力或区分肌肉力不足与腕部跟踪偏差的贡献。track_ok和track_max采用log step 70的服务终结读数（最大误差仅报告至0.001米）；末步track_pos_err未报告数值，保留null。未将agent在log step 81关于腕部质量、刚度和可达性的离线自述当成已核实的环境缺陷。"
+ },
+ "summary_zh": "末次评分episode执行完57个控制步，但苹果始终未随MyoHand完成参考抬升，唯一节点未达成，最终0/1。视频0.36—0.60 s可见手指离开而苹果留在桌面，推断未形成承载抓握；服务在log step 70报告47/57步在轨、最大误差0.070米，均未通过完整轨迹验收。缺少逐步状态与接触力记录，抓取接触设计的归因来自视频和提交脚本的综合推断，末步位置误差数值未知。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "唯一节点未达成、终值0、断点apple_reference_tracked与服务log step 70的判负读数（47/57步在轨、最大误差0.070米）一致；视频中手指仅触碰苹果上部，苹果未被带起，没建立接触/抓取与接触设计的标注可接受，未发现需要改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:51:37Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--yf45bgc/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-apple-lift-i00-standard--yf45bgc/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a4b16e17d0f9b19793ea0c6eb90b521f27a1e233923962f20f4bb3a9b7ebf705",
+   "video_sha256": "89df15ba93889bbcc7d09796bc8d63d5f5abae1d87120bd387a0d667cbd4341f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--2y7srji.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--2y7srji.json
new file mode 100644
index 0000000000000000000000000000000000000000..a72429f3146c90172d74b339a71b0a805df3e1ea
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--2y7srji.json
@@ -0,0 +1,205 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-privileged--2y7srji",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 38,
+   "sim_steps": 38,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 79,
+   "agent_resets": 1802,
+   "agent_sim_steps": 605,
+   "agent_records": 1,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.673,
+   "peak_processes": 18,
+   "peak_disk_gb": 0.989,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-privileged--2y7srji/replay.mp4",
+  "duration_s": 1.56,
+  "fps": 25.0,
+  "frames": 39,
+  "sha256": "28c40be95a5862699276dfc237ab706d150587517b805c2379930697f9c8d935"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.32,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.60 s 的两视角中，MyoHand 手指在兔子一侧接近、轻触后向上离开，未见拇指与其余手指相对包夹兔身；兔子仍靠近桌面。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.64–1.52 s，兔子发生转动，但未被手持续夹持带起；视频 0.96 s 后手指与兔子明显分离，至视频 1.52 s 仍未恢复抓握或完成整段检视运动。"
+     },
+     {
+      "t": 1.52,
+      "log_step": 194,
+      "text_zh": "观察：log step 194 对最终交付轨迹的生成及两次重放均报告 38 步后失败，终止原因为物体距参考或手腕超过 0.25 米，生成状态的 track_ok=19；log step 183 给出 track_steps=99。评分器 reward 同时报 sim_steps=38、task_failure=1、deterministic=1、instance_ok=1，与视频末尾手物分离相符；这里不将控制步数换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–0.60 s，单只 MyoHand 的伸出手指在兔子侧面作轻微拨碰，随后抬离；未见相对夹紧形成抓握。初始同一冻结状态的接触检查（log steps 36、38）亦显示食指、中指末节接触兔身，此检查仅辅助识别初始接触，不代表后续时序。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.8,
+   "log_step": 194,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80 s，MyoHand 已向上移动，兔子未被包夹并随手抬起；视频 0.96–1.52 s 手物分离持续增大，未修复。log step 194 对交付文件的重放确认在第 38 个控制步触发物体距参考或手腕过远的失败终止，未执行完整 99 步。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.60 s 手指虽在兔身一侧拨碰，却未形成拇指与其他手指的相对包夹；视频 0.80–1.52 s 手上移而兔子没有被持续带起，随后手物分离。缺失的是搬动兔子所需的抓握接触，未见先稳定持物再滑脱的阶段。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "推断：视频 0.32、0.60、0.80–1.00 s 可见手指闭合区域未包住兔身，手上移时没有承载兔子；log step 193 的最终生成脚本直接用参考手关节姿态加 6 帧偏移作为 PD 目标并录制，log step 194 确认该文件重复失败。结合这套交付动作的接触形态，所用抓取布局未建立可随手搬动兔子的夹持，是跟踪失败的直接原因；尚未验证可行替代握法。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "未提供状态日志或成功示范视频，无法确定逐帧接触力及精确接触对；inputs/run.json 的 reward 未报告四项原生指标的原始数值，故 native_metrics 留 null。log step 194 仅给出该交付轨迹 track_ok=19，不能补齐末步位移、旋转误差及 track_max。log step 207 关于几何导致所有抓法不可行的说法是 agent 自述，未穷尽验证，不能据此认定任务或环境问题。"
+ },
+ "summary_zh": "交付轨迹尝试用单手拨碰并带动兔子，但没有形成可持续搬动的抓握，兔子未沿参考完成检视运动，节点最终为 0/1。视频约 0.80 s 起手物逐渐分离，log step 194 确认交付轨迹仅执行 38 个控制步即因物体距参考或手腕过远终止，合格采样为 19/99；证据支持抓取和接触设计失败。缺少状态日志、成功示范及评分器原生误差数值，无法进一步确认接触力或替代抓法。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、现象与根因均与视频和日志一致：视频中手指轻触后上移，兔子未被带起而分离；log step 183/194 确认交付轨迹 38 步后失败、track_ok=19。未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:00:43Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--2y7srji/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--2y7srji/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "677a3b0c127dcc74567c7200e7a707af74cc029ed2aac2c68d94f00047fe293d",
+   "video_sha256": "28c40be95a5862699276dfc237ab706d150587517b805c2379930697f9c8d935",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--9eejzm4.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--9eejzm4.json
new file mode 100644
index 0000000000000000000000000000000000000000..aa509562cce30ec2403eeff51ea3c7b8c22b0f49
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--9eejzm4.json
@@ -0,0 +1,205 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-privileged--9eejzm4",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 34,
+   "sim_steps": 34,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 74,
+   "agent_resets": 14862,
+   "agent_sim_steps": 1233,
+   "agent_records": 0,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.174,
+   "peak_processes": 16,
+   "peak_disk_gb": 0.245,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-privileged--9eejzm4/replay.mp4",
+  "duration_s": 1.4,
+  "fps": 25.0,
+  "frames": 35,
+  "sha256": "4189602701e4a3fa6a3fc5d058c2882f56a06d28568eeaaa5f5b837e4a86e13e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–1.36 s 的两个视图；视频 0.64–0.80 s，MyoHand 的手指弯曲并贴近兔子背侧，兔子仍靠近桌面，没有随手被稳定夹持抬起。"
+     },
+     {
+      "t": 1.04,
+      "log_step": null,
+      "text_zh": "观察：视频 0.84–1.36 s，兔子与收拢的手逐渐分离；视频 1.04 s 已可见手指与兔子之间的空隙，随后兔子向画面下方移出，未出现连续持物转向。"
+     },
+     {
+      "t": null,
+      "log_step": 78,
+      "text_zh": "观察：log step 70 运行生成脚本更新交付文件；log step 78 对交付文件作两次重放，均返回 success=false、steps=34、rows=34，原因均为物体距参考或手腕超过 0.25 m；两次末态哈希一致。评分器 reward 同时报告 sim_steps=34、task_failure=1、deterministic=1；完整 99 个控制步的节点验收未完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.64–1.04 s，单只 MyoHand 弯曲的手指在兔子背侧、上侧接触并拨动兔子，兔子发生小幅移动、倾转后与手分离；未见拇指与其余手指形成对向夹持并承载兔子，因此按实际拨动接触记录。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.84,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.80–1.04 s，机器人收拢手指并移动前臂时，兔子未随手被夹持抬起，反而与手逐渐分离；视频 1.04–1.36 s 未恢复有效持物运动。log step 78 中交付文件的两次重放均在第 34 个控制步触发物体距参考或手腕超过 0.25 m 的失败，未执行完整参考运动。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.84,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.80 s 可见手指靠近、接触兔子，但未见对向夹住兔子并带其离开桌面的持物阶段；视频 0.84–1.36 s 手与兔子持续分离。推断：未建立沿参考运动所需的有效抓握接触；这里不表示手指从未碰到兔子。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "推断：视频 0.64–1.04 s 的手指主要在兔子背侧拨动，未形成可承载的对向夹持，说明采用的手形与接触布局未建立有效抓握。日志提供独立支持：log step 19 将参考手形和兔子位姿直接放入运动学模型时，输出多处手指与兔子约 0.02–0.04 m 的穿透；log step 60–61 的会话诊断即使从参考姿态开始并增加屈肌驱动，兔子仍下移、未稳定保持。这些是会话测试，只用于解释接触方案的困难，不充作评分轨迹的节点进度；尚未验证替代抓法。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "未提供状态日志或成功参考视频；两视图分辨率有限，无法确认每根手指的接触对、接触力和短暂遮挡处的接触细节。评分器 reward 未报告 track_pos_err、track_rot_err、track_ok / track_steps、track_max 的原始数值，四项 native_metrics 均留 null；其他会话测试的数值不代替交付轨迹指标。视频时长与 34×0.02 s 不一致，本文时间均直接取已检查的视频帧，未将控制步或 agent 时间换算为视频时间。",
+  "unmet": {
+   "bunny_motion_tracked": false
+  }
+ },
+ "summary_zh": "评分轨迹尝试用单只 MyoHand 带动兔子，但未建立能抬起并连续转向的有效抓握，唯一节点未完成，终态为 0/1。视频 0.84–1.36 s 显示兔子逐渐与手分离，交付文件两次重放均在第 34 个控制步失败；接触布局不适配是有视频和会话诊断支持的根因，可行替代抓法尚未验证。缺少状态日志和跟踪误差原始指标，无法给出精确接触力或容差数值。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频、log 一致：唯一节点未达成，手只在兔子背侧拨动、未形成包握，随后兔子与手分离；log step 78 的两次重放均在第 34 步失败。断点、现象与根因有证据支持，时间均取视频帧。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:01:18Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--9eejzm4/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--9eejzm4/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "28ef106cfad091df9847a88ede8069c68be5d6e62b97686ceb0a4d81156df6c5",
+   "video_sha256": "4189602701e4a3fa6a3fc5d058c2882f56a06d28568eeaaa5f5b837e4a86e13e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--cdagotn.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--cdagotn.json
new file mode 100644
index 0000000000000000000000000000000000000000..24decd5ad71c966865e598f30bb357dc0a50f16e
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--cdagotn.json
@@ -0,0 +1,202 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-privileged--cdagotn",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 40,
+   "sim_steps": 40,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 154,
+   "agent_resets": 280,
+   "agent_sim_steps": 4386,
+   "agent_records": 46,
+   "agent_replays": 8,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.13,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.244,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-privileged--cdagotn/replay.mp4",
+  "duration_s": 1.64,
+  "fps": 25.0,
+  "frames": 41,
+  "sha256": "056a35cd97dfeb56fc0db105fb571656384f93f4b4aab7ff8128bba86ac9be29"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.16 s，MyoHand 指端贴近兔子后侧；视频 0.20–0.64 s，手逐渐离开兔子，兔子仍支承在桌面，未随手抬起。"
+     },
+     {
+      "t": 1.6,
+      "log_step": 169,
+      "text_zh": "观察：逐帧查看视频 0.64–1.60 s，兔子持续留在桌面，空手与兔子分离，未发生抓取后的搬运或连续转向。log step 169 对交付文件的两次重放均返回 success=false、steps=40，原因为物体距参考位置或手腕超过 0.25 m；run.json 的评分器字段 n_actions=40、sim_steps=40、task_failure=1、deterministic=1，与提前结束相符，未完整执行要求的 99 个控制步。视频时刻取自帧，不以 agent 时间或控制步时长换算。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.16 s，单只 MyoHand 的指端贴压兔子后侧、近底缘处，随后手移开；未见拇指与其余手指形成对向夹紧或将兔子托离桌面。具体接触手指和压力大小不可辨。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.24,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.20–0.32 s，MyoHand 向上、向外移开时未带起兔子；视频 0.64–1.60 s 始终空手，兔子留在桌上，该跟踪步骤未被修复。log step 169 的交付轨迹重放在第 40 个控制步因物体距参考位置或手腕过远而失败。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.24,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.16 s 虽有指端贴近、贴压兔子的初始接触，但视频 0.20–0.64 s 手移开时兔子不随手运动，视频 0.64–1.60 s 兔子仍在桌面且手中为空。推断：该步骤需要的承载抓握从未建立，没有可见的先持物、后滑落过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.24,
+    "log_step": 162,
+    "evidence_zh": "推断：所用手指与兔子的接触布局未形成承载包围。视频 0.00–0.32 s 可见手指仅在兔子一侧附近活动，手离开时未夹住物体，视频 1.00–1.60 s 手指在空处弯曲；log step 162 的生成脚本采用预抬腕、随后追踪参考手指姿态的固定阶段方案，log step 164 修复变量后保存该方案，log step 169 验证了最终交付文件的失败重放。现有证据支持抓握接触方案未奏效，未验证有可行替代握法。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "无状态日志或成功示范，不能确认具体接触指对及接触力；初始贴压的细部受兔子遮挡。评分器 reward 未报告四项原生跟踪指标，全部保留 null；会话中打印的 oe 是绝对目标距离，不能冒充相对位移跟踪误差。agent 在 log step 155、176 自述参考兔子初始高度存在几何偏差，该说法不足以排除所有合法抓取方案，环境根因待核实。"
+ },
+ "summary_zh": "MyoHand 在兔子旁有指端贴压和收指、抬手动作，但未形成能带起兔子的抓握；兔子全程留在桌面，沿参考运动节点最终 0/1。交付的 40 步轨迹经确定性重放仍提前触发物体距离失败，断点在跟踪操作中的抓握建立阶段；接触布局未形成承载抓握，无状态日志及原生误差数值可进一步核验。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点、现象与时钟均与视频和 log 相符；兔子全程留在桌面，手空手移开，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:59:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--cdagotn/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--cdagotn/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "00f2cbcd40b1d2d44646d5139fe761a26d8163c1697286d1869fa8e506e25d2c",
+   "video_sha256": "056a35cd97dfeb56fc0db105fb571656384f93f4b4aab7ff8128bba86ac9be29",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--kwqryxj.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--kwqryxj.json
new file mode 100644
index 0000000000000000000000000000000000000000..7b52a348591bd0a424cccd3eb31753cfd7f093b3
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--kwqryxj.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-privileged--kwqryxj",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 41,
+   "sim_steps": 41,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 56,
+   "agent_resets": 111,
+   "agent_sim_steps": 1586,
+   "agent_records": 0,
+   "agent_replays": 9,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.652,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.239,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-privileged--kwqryxj/replay.mp4",
+  "duration_s": 1.68,
+  "fps": 25.0,
+  "frames": 42,
+  "sha256": "810d8dafdb3207de51e6970fe1612d1b2104cc92086c93f90916b99297ddb930"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的两个视角显示兔子仍由桌面承托，MyoHand 单手位于兔子后侧；视频 0.00–0.64 s 手指靠近兔子，未见兔子被稳定抬离桌面。"
+     },
+     {
+      "t": 0.88,
+      "log_step": null,
+      "text_zh": "观察：逐帧查看视频 0.68–1.04 s，MyoHand 手指贴近、推触兔子后部，兔子向前倾转并向右侧视角的左下方偏移；视频 1.12–1.64 s 兔子与手部继续分离，片尾没有形成随手抬起并连续转向的运动。"
+     },
+     {
+      "t": null,
+      "log_step": 76,
+      "text_zh": "观察：log step 72 保存的最终候选为41步动作；log step 73、76 对该交付文件进行两次官方重放，均报告 steps=41、success=false、相同 state_sha=e51782546d174e94，reason 为兔子距参考或手腕超过0.25米；inputs/run.json 的评分字段也为 n_actions=41、sim_steps=41、replay_success=0、deterministic=1。"
+     },
+     {
+      "t": 1.64,
+      "log_step": null,
+      "text_zh": "推断：视频末帧兔子未被手稳定带起，且交付回放在完成要求的99个控制步之前终止，因此完整参考运动节点从未达成；不能将初段误差较小的若干采样当作整个事件完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.64–0.92 s，MyoHand 单手的手指在兔子后侧推触，兔子随之倾转、偏移；两个视角未显示拇指与其余手指形成对向夹持并稳定带起兔子，因此实际接触记为推/扫。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.8,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80 s 起，MyoHand 仍在操作兔子，但兔子开始明显倾转、偏移；视频 0.88–1.64 s 兔子离开手部，没有被稳定带起，后续未修复。log step 76 对最终41步交付文件的两次官方重放均提前失败，未完成99步参考运动。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.88,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.72–1.04 s，兔子在MyoHand 手指推触下前倾、转倒并向画面左下方移动；视频 1.12–1.64 s 手部继续运动，而兔子未随手稳定抬升，位置与朝向明显失去所需的受控检视运动。未见已经稳定持有后再滑脱的过程，因此不标没保持住。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 61、67 的控制器用参考关节姿态的PD误差和肌肉力矩分配生成动作，log step 72 选定的最终参数仍只得到41步后终止，log step 76 在新进程中重复同一失败；结合视频 0.64–1.64 s 手指推触后兔子倾转并离开手部，说明这套肌肉与前臂控制未实现稳定接触下的物体运动。证据支持控制失败，但不足以确定是接触几何、轨迹跟踪还是反馈控制的具体子类。"
+   }
+  ],
+  "gap_zh": "无状态日志、成功示范或本次交付的完整逐步位姿及接触力数据，无法精确核对每步误差和接触分布；评分 reward 未报告 DAG 的四项原生指标，故其 value 均留 null。log step 77 关于参考姿态物理不一致的说法是 agent 的解释，未证明所有合法控制方案均不可行，不据此标任务或环境根因。控制步的仿真时长与视频播放时长不同，本文件的时间锚点均直接取自视频帧，未用 agent 时间或控制步数换算。"
+ },
+ "summary_zh": "评分片段中MyoHand 单手推触兔子，但未将其稳定带起；完整参考运动节点未完成，终态为0/1。视频约 0.80–1.04 s 兔子倾转并偏离手部，之后未恢复，最终41步交付在两次官方重放中均提前终止；现象为位姿不准，根因归为控制。缺少状态日志和完整原生误差数据，无法进一步区分接触设计与轨迹、反馈控制问题。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点和现象/根因与视频及log一致；视频时间均取自帧，log step id存在，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:06:14Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--kwqryxj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--kwqryxj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "11186c048b71e1e2a804291be7ff484d65eff1f4532fa53bc3187d04bc50fc4d",
+   "video_sha256": "810d8dafdb3207de51e6970fe1612d1b2104cc92086c93f90916b99297ddb930",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw.json
new file mode 100644
index 0000000000000000000000000000000000000000..32bfea14ce3b0cedf47980bd19ef7ee6f9f3929b
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw.json
@@ -0,0 +1,226 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-privileged--mnedsnw",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 41,
+   "sim_steps": 41,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 138,
+   "agent_resets": 4910,
+   "agent_sim_steps": 59453,
+   "agent_records": 0,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.614,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.842,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw/replay.mp4",
+  "duration_s": 1.68,
+  "fps": 25.0,
+  "frames": 42,
+  "sha256": "c09c425515c6a3aa65dc1380199a95e59b4e1d9fb0ba8b01941b1070842a4ac0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频 0.00–1.64 s 的全部42帧及侧视放大图；MyoHand 手指在兔子后侧张合，视频 0.80–1.64 s 兔子仍贴近桌面，后段仅有倾转，没有随手完成抬起并连续转向的运动。"
+     },
+     {
+      "t": null,
+      "log_step": 165,
+      "text_zh": "观察：log step 164 运行交付生成器并对同一交付文件重放两次；log step 165 返回两次均为41步、同一末态哈希、success=false，终止原因均为物体距参考或手腕超过0.25米。该交付生成过程末步 track_pos_err=0.27266米、track_rot_err=2.2864弧度、track_ok=19，要求的 track_steps=99，故完整跟踪节点未达到。"
+     },
+     {
+      "t": 0.8,
+      "log_step": 8,
+      "text_zh": "推断：log step 8 所示 render_video 在初态后每执行一个动作写一帧，帧率25；交付的41步对应视频末帧 1.64 s，控制第20步对应视频 0.80 s，已在该帧核对兔子未随手抬起。因此不能用控制步长0.02秒直接作为视频时刻。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "推断：视频 0.00–0.20 s 及 1.20–1.64 s 单只 MyoHand 的手指贴近并擦压兔子后侧，后段兔子有倾转；侧视未见手指与拇指从两侧夹住兔子，也未见承重抬起，按可见擦压接触记录。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.8,
+   "log_step": 165,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.80–1.64 s，MyoHand 继续屈指和转腕，兔子留在桌面附近而未被带起完成检视运动。log step 165 对交付文件的生成检查显示控制第20步位置误差0.03494米，之后第30步增至0.18372米；两次重放均在第41步失败终止，后续未修复此步骤。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": 165,
+    "evidence_zh": "推断：逐帧检查视频 0.00–1.64 s 的双视角及侧视放大图，MyoHand 手指在兔子后侧张合并擦压，未见拇指与其余手指从两侧夹住兔子；视频 0.80–1.64 s 兔子仍贴近桌面，后段仅倾转，未随手承重抬起。据此判断所需的有效抓握接触未建立，非指完全没有触碰。log step 165 的交付生成检查中，控制第20、30、41步位置误差分别为0.03494、0.18372、0.27266米，track_ok 停在19，仅佐证操纵未成功；早期遮挡下是否有短暂对向夹持仍不确定。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 19 的交付生成器以逆动力学和肌肉分配跟踪手部参考，log step 83 显示其仍采用手部位姿误差控制；log step 164–165 重新生成并重放该文件时，手部误差范数从控制第10步约0.98增至第30步约2.99，物体跟踪合格计数停在19。结合视频 0.80–1.64 s 手指运动却未将兔子带起，说明该控制未能形成并维持有效的物体操纵；没有证据把问题进一步确定为接触设计、逆解或某个参数。"
+   }
+  ],
+  "gap_zh": "没有状态日志和成功参考视频；双视角能判定未完成抬起跟踪，但兔子背后部分手指接触被遮挡，无法确认早期是否短暂形成对向夹持、具体接触力及控制子类。run.json 的 reward 未报告 DAG 四项原生指标，故 native_metrics 均为 null；log step 165 只有交付生成检查的部分舍入数值，未提供完整误差序列和 track_max。"
+ },
+ "summary_zh": "交付轨迹只执行41个控制步，兔子没有随 MyoHand 完成抬起和连续转向，唯一跟踪节点未达到，终态进度0/1。视频可见手指在兔子后侧擦压而未形成承重抬起，据此推断现象为未建立有效抓握接触，根因保留控制未实现有效操纵；缺少状态日志和完整误差序列，不能细分控制根因或确认早期遮挡下的短暂夹持。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "时钟、节点（未达成）、断点 bunny_motion_tracked 和 contact_actual 均无误；问题在现象：'运动·位姿不准' 只是复述“兔子未被带起/未跟上”这一未达指标（N10），而同一文件自己的证据显示手指从未从两侧环抱兔子，更贴合 '接触·没建立接触'。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 bunny_motion_tracked、时钟映射（41步=1.64 s）、contact_actual 压/擦 与第1轮修改后的现象（没建立接触，标为推断）均与视频及 log step 165 相符，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:13:36Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--mnedsnw/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "abbfdeb83389b987d29864d4fe65591d2ec579046c802277bb808027eb6e3f69",
+   "video_sha256": "c09c425515c6a3aa65dc1380199a95e59b4e1d9fb0ba8b01941b1070842a4ac0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--ngyptxu.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--ngyptxu.json
new file mode 100644
index 0000000000000000000000000000000000000000..0f22155caddd472bf49116ca6663771a2b4c37e1
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-privileged--ngyptxu.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-privileged--ngyptxu",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 38,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 239,
+   "agent_resets": 181,
+   "agent_sim_steps": 2651,
+   "agent_records": 1,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.086,
+   "peak_processes": 7,
+   "peak_disk_gb": 0.265,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-privileged--ngyptxu/replay.mp4",
+  "duration_s": 1.56,
+  "fps": 25.0,
+  "frames": 39,
+  "sha256": "ac27d1d55f53069d7c59664db8b518f0e66ef216c1c6d3a92af31e7db7de231b"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.4,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.72 s 的两个并排视角中，MyoHand 单手向白色兔子靠近，手指碰动兔子，兔子仍在桌面附近，未见被对向夹住并随手抬起。"
+     },
+     {
+      "t": 0.88,
+      "log_step": null,
+      "text_zh": "观察：视频 0.76–1.20 s，兔子沿桌面移动、转偏，手指在其右侧收拢并逐渐离开；视频 1.24–1.52 s，手与兔子已分开，兔子留在桌面，未完成抬起后的连续检视运动。"
+     },
+     {
+      "t": null,
+      "log_step": 298,
+      "text_zh": "观察：log step 298 的最终生成脚本输出 Session 状态：steps=38、track_steps=99、track_ok=18、track_pos_err=0.27568 m、track_rot_err=2.0843 rad，done=true，原因为物体距参考位置或手腕超过 0.25 m；保存的是 99 行动作。"
+     },
+     {
+      "t": null,
+      "log_step": 304,
+      "text_zh": "观察：log step 304 对最终保存文件执行两次官方重放，均只执行 38/99 行、success=false，末态哈希均为 2f5854fe8149468c，与 log step 298 一致；run.json 的 reward 也报告 sim_steps=38、task_failure=1、deterministic=1。结合视频中兔子未被带离桌面的结果，整段运动节点从未完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.32–1.12 s，MyoHand 单手的手指从兔子一侧接触并拨推，使兔子在桌面附近移位、转偏；随后手离开。未见拇指与其余手指形成包围兔子的对向夹持，故不记为抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.8,
+   "log_step": 298,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.76–0.92 s，MyoHand 手指收拢、手臂开始移走时，兔子仍在桌面附近被拨动，未随手抬起；视频 1.24–1.52 s 两者分离，未修复。log step 298、304 记录最终文件均在第 38 个控制步因物体远离参考或手腕而终止，完整 99 步运动未完成；此处视频锚点直接取自画面，未用 agent 工作时间换算。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.64–1.12 s，机器人手指靠着兔子一侧收拢、拨动兔子，随后手向右上方移开；视频 1.24–1.52 s 兔子仍留桌面，手已空手离开。推断：实际有碰触，但未建立能夹持并带动兔子抬起的抓持接触；没有先持稳再滑落的可见阶段。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 297,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：机器人未能通过所生成的腕部与肌肉命令建立有效抓持。log step 297 的最终生成器仅对前六个基座关节计算跟踪反馈，对手指屈肌、拇指肌及手内肌按控制步数统一递增激活，未根据兔子相对手的位置或是否已抓住进行修正；log step 298、304 证实该文件重放失败，视频 0.76–1.52 s 可见手收拢后离开而兔子仍在桌面。现有证据不能进一步区分接触几何、肌肉协同和腕部跟踪各自的贡献，归于控制总类。"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功参考视频；早期部分手指被兔子遮挡，不能逐对确认手指接触与接触力，也不能精确分离抓取设计和跟踪控制的贡献。run.json 的 reward 未报告 DAG 的四项原生连续指标，native_metrics 的值均留 null；log step 298 的 Session 末态读数已单独列为日志证据，不能据此补出全程最大误差 track_max 或完整合格采样序列。视频仅 39 帧、25 fps，不把每帧时长等同于任务的 20 毫秒控制步，所有视频锚点均直接读帧。"
+ },
+ "summary_zh": "MyoHand 单手接触并拨推兔子，但兔子未被有效夹持、抬起，完整沿参考运动节点未达成，最终进度为 0/1。视频约 0.80 s 开始清楚显示兔子留在桌面而手逐渐离开，最终文件虽含 99 行动作，官方重放均在第 38 个控制步失败；现象为未建立有效抓持，根因归于控制。无状态日志或成功示范，评分汇总也未给出原生连续指标，具体控制子因及精确接触力不能确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点未达成、断点在 bunny_motion_tracked、现象（没建立接触）与根因（控制，未细分）均有视频帧和 log step 298/304 的支持；视频时间均为视频钟，未见换算或越界。无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:02:10Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--ngyptxu/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-privileged--ngyptxu/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "13f1b9d9d03a025dad590224bf5b8d632555400f334ccb9b6c0ac4fb6b74f859",
+   "video_sha256": "ac27d1d55f53069d7c59664db8b518f0e66ef216c1c6d3a92af31e7db7de231b",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--beyyxma.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--beyyxma.json
new file mode 100644
index 0000000000000000000000000000000000000000..eb7c424c21e79e338833ee8336c092a9d7f3e842
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--beyyxma.json
@@ -0,0 +1,237 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-standard--beyyxma",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 10,
+   "sim_steps": 10,
+   "physics_steps": 100,
+   "steps": 10,
+   "calls": 2,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 10,
+   "elapsed_s": 932.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.603,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.372,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-standard--beyyxma/replay.mp4",
+  "duration_s": 0.44,
+  "fps": 25.0,
+  "frames": 11,
+  "sha256": "fe6db030e566ca1ea78aceca8852813f9b0041f62a1d213996bfc44df7a7ece6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 30,
+      "text_zh": "观察：已逐帧检查视频 0.00–0.40 s 的全部11帧；MyoHand 的手指起初靠近桌面兔子，随后弯曲并撤离，兔子始终留在桌面，未见被手抬起或连续转向。"
+     },
+     {
+      "t": 0.4,
+      "log_step": 30,
+      "text_zh": "观察：log step 30 中 agent 向现场服务提交的是对照本地副本的30行校准动作（前臂与肌肉使用预设正弦模式），服务返回 executed rows=10、done=True，reason 为兔子距参考或腕超过0.25米；评分器 reward 的 steps=10、episodes=1、resets=0、success=0。该局提前终止，没有执行节点要求的完整99步。"
+     },
+     {
+      "t": null,
+      "log_step": 83,
+      "text_zh": "推断：完整日志中 log step 30 是唯一现场动作提交，此后直到 log step 83 均为本地建模、控制器修改与测试；run.json 的 sim_total.submit=1、resets=0 与之吻合。因此视频所示校准局就是评分局，后续本地测试不构成节点进展；本节点的参考跟踪工作未启动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.0,
+   "log_step": 30,
+   "where": "首步前",
+   "evidence_zh": "推断：视频 0.00–0.40 s 仅见 MyoHand 撤开、兔子留桌，未进入抬起并跟随参考的动作；log step 30 的提交代码明确用于校准现场与本地副本，服务在第10个控制步报失败。之后无现场重置或提交（完整日志至 log step 83；reward resets=0、steps=10），参考跟踪步骤始终未开始。这里的视频锚点取起始帧，不将控制步的仿真时间直接换算为视频时间。"
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 30,
+    "evidence_zh": "观察：视频 0.00–0.40 s 的手逐渐离开兔子，兔子未离开桌面；log step 30 只提交用于验证控制接口的正弦校准模式，未提交抓持兔子后沿参考运动的控制序列，服务随即报 done=True。评分局仅有该次校准，后续 log step 31–83 没有新的现场动作。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent 未能调出可执行的兔子抓持与跟踪控制器。作为会话证据，log step 69、73 的本地控制器测试均提前触发终止，兔子质心高度约维持在0.046米；修改前臂阻尼与补偿后的 log step 82 仍在本地 rollout 的第39个索引处终止，输出的兔子质心采样仍约0.046米高、没有随升高的参考抬起。log step 31 对本地副本与现场校准所得关节位置、速度及肌肉激活逐步比对吻合，为使用这些控制调试结果解释未形成任务控制提供支持；无法进一步确定是抓取设计还是轨迹跟踪问题。"
+   },
+   {
+    "cat": "交付",
+    "item": "交付操作出错",
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent 将接口校准测试局留作评分局，属于有证据的交付选择操作出错。inputs/run.json 的 log step 30 代码向现场服务提交30行测试动作，前臂使用正弦模式，肌肉列含 A[:,6]=2.0、A[:,7]=-1.0；服务返回 control_steps_chunk=10、executed rows=10、done=True，reason 为物体距参考或腕超过0.25米。task_text 的 Episodes and resets 规则明确允许10次 reset，且只有最后一个 episode 被评分；但 log step 31–83 无现场 reset 或 submit，sim_total 为 submit=1、reset=0，reward 为 episodes=1、resets=0、reset_budget=10、steps=10。因此这次提前终止的校准局成为唯一评分局，后续本地控制器测试没有替换它；视频 0.00–0.40 s 的全部11帧仅见手撤开、兔子留桌，与该局未开展参考跟踪相符。"
+   }
+  ],
+  "gap_zh": "没有状态日志、接触对或成功示范；视频起始手指与兔子邻近处有遮挡，不能确认短暂相贴，更不能据此认定已抓握。评分局只有校准动作，未对参考跟踪步骤作实际任务尝试，因此未列该节点的 contact_actual。评分器未报告 track_pos_err、track_rot_err、track_ok / track_steps、track_max，四项指标均为 null；本地测试中的误差数值不是评分局指标。一般控制根因依据会话中的本地控制调试，其具体子类与评分局精确接触过程无法确定；交付操作出错另据现场校准提交、任务的最后局评分规则及未重置或再提交的记录确定，不把本地测试当作评分局进展。",
+  "unmet": {
+   "bunny_motion_tracked": false
+  }
+ },
+ "summary_zh": "评分局仅执行了接口校准动作，MyoHand 撤开而兔子始终留在桌面；现场服务在第10个控制步报距离终止，唯一节点“兔子沿参考运动”未达成（0/1），断点在首步前。agent 随后一直在本地调试，测试反复未能抬起兔子，支持会话层面的一般控制根因；同时，任务允许10次 reset 且只评最后一局，agent 却未重置或再提交，使已终止的校准测试局成为评分局，支持交付操作出错。无状态日志、明确接触对及评分器原始跟踪误差，起始短暂接触与控制根因子类不能确定。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达、断点首步前/没尝试均成立（视频 0–0.44 s 仅见手指弯曲撤开、兔子留在桌面；run.json 唯一一次 submit 为 log step 30 的校准块，服务 done=True 于第10控制步）。但缺少一项有证据的根因：agent 把唯一的现场提交用在带极端值（A[:,6]=2.0、A[:,7]=-1.0 等）的测试块上，episode 随即终止，task_text 明示有 10 次 reset 且'只有最后一个 episode 被评分'，agent 却从不 reset、也不再提交，导致评分局就是这次测试块。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达、断点首步前/没尝试成立：视频仅见手指弯曲撤开、兔子留在桌面；唯一一次现场提交是 log step 30 的校准块，服务第10控制步 done=True，sim_total submit=1/reset=0。控制根因明确标为会话证据，交付操作出错有现场提交与最后局评分规则支持。未见需改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:15:34Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--beyyxma/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--beyyxma/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--beyyxma/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--beyyxma/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "66c06a586426ce4ffaaae479d0ffcd86c9ea430039debd7e7a8dc74d7aea3c9d",
+   "video_sha256": "fe6db030e566ca1ea78aceca8852813f9b0041f62a1d213996bfc44df7a7ece6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--ktcdgn5.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--ktcdgn5.json
new file mode 100644
index 0000000000000000000000000000000000000000..d0f55e84392350d9093e86698c084bb614d77ba5
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--ktcdgn5.json
@@ -0,0 +1,214 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-standard--ktcdgn5",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 39,
+   "sim_steps": 39,
+   "physics_steps": 390,
+   "steps": 307,
+   "calls": 169,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 307,
+   "elapsed_s": 1363.7,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.611,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.851,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-standard--ktcdgn5/replay.mp4",
+  "duration_s": 1.6,
+  "fps": 25.0,
+  "frames": 40,
+  "sha256": "a8ddde0b914f24793aa59578784cf25b8e19f1b695c14e13d1ef9745aa7a5b85"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.32,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.32 s，MyoHand 的手指由张开变为在兔子头部旁合拢，兔子仍由桌面支撑，未被拇指与其余手指包夹；视频 0.36–1.56 s 也没有随手抬起并连续转向的运动。"
+     },
+     {
+      "t": 1.56,
+      "log_step": 87,
+      "text_zh": "观察：视频 1.20–1.56 s，合拢的手指碰推兔子头部，兔子在桌面上偏转；log step 83 启动最后一次 reset 后的评分局，log step 87 的机器人服务状态为 episode=11、control_steps=39、done=True，原因为兔子距参考或手腕超过 0.25 米；评分器记录 n_actions=39、live_success=0。该局未执行完任务要求的 99 个控制步。视频时间直接来自帧，不将控制步或 agent 时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 1.20–1.56 s，MyoHand 已合拢的手指从侧上方碰推兔子头部，兔子在桌面上偏转；此前手指在兔子旁空合拢，未见包夹兔子的对向抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.32,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.32 s，MyoHand 已在兔子旁合指，兔子没有进入指间；到视频 1.56 s 仍未被抬起，后段只有碰推与桌面偏转。log step 87 的最终服务状态确认评分局在第 39 个控制步因兔子距参考或手腕超过 0.25 米而结束，未完成整段参考运动。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.32,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.20–0.76 s，MyoHand 的手指在兔子头部旁闭合，兔子不在指间且仍在桌面；视频 0.80–1.56 s 手指逐渐碰到兔子头部并使其偏转，却始终没有包夹抬起。推断：虽有后段碰推，沿轨迹搬运所需的抓握接触从未建立，不是先持住再滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.32,
+    "log_step": 87,
+    "evidence_zh": "推断：视频 0.20–0.76 s 显示手指在兔子旁提前合拢，随后保持合拢姿态逼近头部，视频 1.20–1.56 s 只以手指外侧碰推，接触几何没有形成能承载兔子的对向包夹。log step 87 展示的最终控制器对 FDS/FDP、FPL、OP 肌肉设置闭合激励，并在后段继续发送闭指动作，和视频中的闭指接近一致；该抓取执行没有建立搬运所需的承载布局。没有已验证的替代抓法。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "无状态日志和成功参考视频，缺少接触对、接触力及物体位姿，不能判定具体是哪根手指接触或量化承载力。评分器与服务输出未报告 track_pos_err、track_rot_err、track_ok / track_steps、track_max，四项原生指标均为 null；终止提示将距参考与距手腕合并，无法区分触发项。日志中最终控制器的中段源码被截断，无法进一步核实全部闭指与接近时序；未推定可行替代抓法。"
+ },
+ "summary_zh": "评分局中，MyoHand 在兔子旁合指，随后碰推兔子头部，始终未将兔子抓起沿参考运动，唯一节点未达成（0/1）。接触几何未形成承载抓握；log step 87 的服务状态显示最后一局在第 39 个控制步因兔子距参考或手腕超过 0.25 米而提前终止。无状态日志、成功参考视频及原始跟踪误差数值，无法量化跟踪精度或区分终止提示中的两种距离条件。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、没建立接触与抓取和接触设计根因均有帧和服务读数支持，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:01:49Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--ktcdgn5/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--ktcdgn5/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "e603a9c84e012642c3305d180de092b88da9bd5cbaa0250d994c3d90d4ae78e8",
+   "video_sha256": "a8ddde0b914f24793aa59578784cf25b8e19f1b695c14e13d1ef9745aa7a5b85",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--kzaze3u.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--kzaze3u.json
new file mode 100644
index 0000000000000000000000000000000000000000..284d7a1a7dccc0180a2142942956161acbd205f6
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--kzaze3u.json
@@ -0,0 +1,217 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-standard--kzaze3u",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 12,
+   "sim_steps": 12,
+   "physics_steps": 120,
+   "steps": 155,
+   "calls": 40,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 155,
+   "elapsed_s": 1376.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.374,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.542,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-standard--kzaze3u/replay.mp4",
+  "duration_s": 0.52,
+  "fps": 25.0,
+  "frames": 13,
+  "sha256": "0aae0d942e893992ead67877ff737a30c884c33acb60f170d79e0fbe5a28fbf9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.16,
+      "log_step": 51,
+      "text_zh": "观察：已查看视频 0.00–0.48 s 的全部13帧；MyoHand 手指从兔子旁的张开姿态逐渐闭合，视频 0.16–0.48 s 手与兔子间的间隙增大，兔子一直留在桌面，没有随手抬起和连续转向。"
+     },
+     {
+      "t": 0.48,
+      "log_step": 51,
+      "text_zh": "观察：log step 51 中机器人服务返回 episode=11、control_steps=12、done=true，终止原因为物体距参考或手腕超过0.25米；评分器 reward 同时记录 graded_by_live=1、n_actions=12、sim_steps=12、live_success=0。log step 52 的重置失败，log step 55 的最终报告也明确本次评分采用该局。"
+     },
+     {
+      "t": 0.48,
+      "log_step": 51,
+      "text_zh": "推断：评分局只执行12个控制步，且画面未出现兔子被抓起，故未满足完整99步的沿参考运动节点；此判断依据动作与提前终止记录，并非仅由零分反推。控制步数仅用于判断执行是否完整，未将控制步或 agent 时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频 0.08–0.48 s，MyoHand 单手在兔子旁屈指闭合并移离，兔子未进入手指的包围中，未被夹持或带起。候选方式为“空抓”：这是朝兔子的抓取尝试，不能标作无需接触的步骤；现有抓握、推/扫、压/擦等方式均不能描述这次未建立目标握持的动作。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.16,
+   "log_step": 51,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.16 s，MyoHand 已开始在兔子旁闭合手指，却没有包住兔子；视频 0.20–0.48 s 手继续移离，兔子仍在桌面，整个评分局没有修复抓取。log step 51 的机器人服务随后报告最后一局在12个控制步后因物体距参考或手腕过远而终止。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.16,
+    "log_step": 51,
+    "evidence_zh": "观察：视频 0.00–0.48 s，MyoHand 手指在兔子旁逐渐闭合，兔子没有被相对手指夹住，闭合后的手移离而兔子留在原桌面区域；本次未建立带动物体所需的抓握接触，未见先持有再脱落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "推断：log step 51 实际执行的脚本从首行即对 FDS/FDP/FPL/OP 等屈肌下发2.0闭合指令，同时将前臂平移目标设为起始关节值加提前20个控制步的参考位移，没有把兔子先移入手指包围区的接近段；视频 0.08–0.48 s 显示手指在兔子旁闭合后移离，未形成相对夹持，支持抓取几何不适配。agent 在 log step 55 也自述纯屈指无法接触兔子，并提出先平移手的修正，但 log step 53 写出的修正版未运行，不能据此确认可行替代。",
+    "source_zh": "日志 / 视频 / agent 自述"
+   }
+  ],
+  "gap_zh": "无模拟器状态日志或成功参考视频；现有双视角足以看出未形成握持，但无法排除起始瞬间的轻微擦碰，不能给出接触力。评分器 reward 未报告 track_pos_err、track_rot_err、track_ok / track_steps、track_max 的数值，四项原生指标均留空；终止提示将距参考和距手腕两种条件合并，不能确定具体是哪一个触发。视频时长与控制步数不直接对应，未建立两者时间映射，全部视频锚点直接取自所查看帧。"
+ },
+ "summary_zh": "评分采用最后一局（episode 11）：MyoHand 屈指抓取时未将兔子包住，兔子始终留在桌面，唯一的沿参考运动节点未完成（0/1），机器人服务在12个控制步后报告距离过远并终止。断点为该节点的操作中，画面和实际执行脚本支持抓取与接触设计不适配；随后编写的修正版未执行，不计入进度。无状态日志和成功参考视频，四项精确跟踪指标未报告，距离终止的具体分支也无法确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频、日志一致：评分局（episode 11）仅12个控制步即终止，兔子始终留在桌面，节点未达成，断点、没建立接触、抓取和接触设计均有证据支持，未转换 agent 时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:03:39Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--kzaze3u/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--kzaze3u/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "8a6a105c2e04cb68e9e0be70f7535575d4650fe00320bd99556629545c62b20b",
+   "video_sha256": "0aae0d942e893992ead67877ff737a30c884c33acb60f170d79e0fbe5a28fbf9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--mt7fuxf.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--mt7fuxf.json
new file mode 100644
index 0000000000000000000000000000000000000000..90424cf62fe0b75449c22e5c7fd30406183855ef
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--mt7fuxf.json
@@ -0,0 +1,213 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-standard--mt7fuxf",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 39,
+   "sim_steps": 39,
+   "physics_steps": 390,
+   "steps": 349,
+   "calls": 348,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 349,
+   "elapsed_s": 1234.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.358,
+   "peak_processes": 8,
+   "peak_disk_gb": 0.274,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-standard--mt7fuxf/replay.mp4",
+  "duration_s": 1.6,
+  "fps": 25.0,
+  "frames": 40,
+  "sha256": "435f2fb2c1b186849b855833f743438b15f0c76ca5c3769dcff9ad717b8be347"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–1.56 s；MyoHand 从兔子后上方靠近并弯曲手指，视频约 0.56–1.00 s 扫碰兔子，兔子在桌面倾斜转动，未见其被包夹并随手抬起。"
+     },
+     {
+      "t": 1.56,
+      "log_step": 50,
+      "text_zh": "观察：视频 1.04–1.56 s，手指闭合在兔子后上方，兔子仍在桌面，与手逐渐分离。log step 50 的服务状态为 episode=11、control_steps=39、done=True，终止原因是物体距参考或腕超过 0.25 m；reward 的 graded_by_live=1、n_actions=39、episodes=11 确认采用最后一局，未执行完整 99 步。视频时间直接取帧，不把服务控制步数或 agent 时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频约 0.56–1.00 s，MyoHand 单手的手指从兔子后上方扫碰其上部，兔子随之在桌面倾斜、转动；随后手指在兔子旁闭合，未形成包夹抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.8,
+   "log_step": 50,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.64–1.04 s，MyoHand 弯指靠近兔子，但兔子向桌面倾斜，未随手形成受控抬升；视频 1.04–1.56 s 仍未补成抓握。log step 50 中最后一局服务在第 39 个控制步提前判负，兔子沿参考运动节点未完成。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": 50,
+    "evidence_zh": "观察：视频 0.56–1.08 s，手指扫碰兔子使其倾斜，随后空合在兔子后上方；视频 1.08–1.56 s，兔子留在桌面而手在其旁移动，没有能承载兔子的对向抓握接触。这里未建立的是抓持接触，并非从未碰到物体。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 0.8,
+    "log_step": 50,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 44 的执行脚本按预定 tclose 开始屈指，并持续执行前臂目标，未依据物体是否抓住调整动作；log step 50 将 tclose 设为 13 并运行最终局，脚本打印的前臂实际关节位置减目标位置，在循环索引 12 的平移分量为约 +0.025、−0.044、−0.008 m，转动分量之一约 +0.079 rad。结合视频 0.64–1.56 s 手指在兔子旁闭合而未带起兔子，说明最终控制未把靠近、屈指和持物运动协调成有效抓取；证据不足以细分接触几何与轨迹跟踪的作用。"
+   }
+  ],
+  "gap_zh": "无状态日志、接触对或力读数，无法确定各手指的精确接触点与抓取失效的控制子类；四项原生跟踪指标在 reward 中均未报告，故 value 为 null。没有成功示范，不能进行参考视频对照；本次未完成由视频中的未抓起及最后一局提前终止共同确定，不从零分单独推断。视频时长 1.60 s 与服务 39 个 20 毫秒控制步的仿真时长不同，本标注全部时间使用直接检查的帧时间，未建立两者映射。"
+ },
+ "summary_zh": "评分采用第 11 局现场结果，唯一的“兔子沿参考运动”节点未完成，最终进度 0/1。视频约 0.64–1.04 s，MyoHand 扫碰兔子后未建立抓握，兔子留在桌面而手指在其旁闭合；log step 50 的服务在第 39 个控制步因物体远离参考或手腕提前终止。控制未实现有效抓取，缺少状态、接触力和原生误差指标，不能进一步确定控制子类或精确误差。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注与视频和日志一致：唯一节点未完成，最终 0/1；视频 1.6 s 内手扫碰兔子后在旁边空合，兔子留在桌面；log step 50 显示第 39 个控制步终止。现象与停在“控制”的根因有证据支持，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:02:38Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--mt7fuxf/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--mt7fuxf/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0082a1d6bf6c8e9e15dc579b151893e4a69768e87f16bf9b28fd732c5b936d64",
+   "video_sha256": "435f2fb2c1b186849b855833f743438b15f0c76ca5c3769dcff9ad717b8be347",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--rtry8gj.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--rtry8gj.json
new file mode 100644
index 0000000000000000000000000000000000000000..bb3773ab6f2b51e783d32446ce7edaeee112fcb6
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--rtry8gj.json
@@ -0,0 +1,221 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-standard--rtry8gj",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 35,
+   "sim_steps": 35,
+   "physics_steps": 350,
+   "steps": 274,
+   "calls": 70,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 274,
+   "elapsed_s": 1925.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.799,
+   "peak_processes": 13,
+   "peak_disk_gb": 1.561,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-standard--rtry8gj/replay.mp4",
+  "duration_s": 1.44,
+  "fps": 25.0,
+  "frames": 36,
+  "sha256": "be54a7f4f4234261d598b0f2d5ba12ffc19635bb1c16fa54431e3086d50c1b8c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的两个视角中，白色兔子在桌面，MyoHand 的手指位于兔子旁上方；此时没有完成整段参考运动。"
+     },
+     {
+      "t": 0.88,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–1.40 s；视频约 0.80–1.16 s 机器人手指收拢，兔子在桌面倾斜，未见兔子被手包夹后随手抬起、连续转向；视频 1.40 s 手与兔子分离，兔子仍在桌面。"
+     },
+     {
+      "t": null,
+      "log_step": 212,
+      "text_zh": "观察：log step 204 启动 g1 伸手闭指局，log step 205 返回控制步 35 提前失败；log step 212 的机器人服务状态确认 episode=11、control_steps=35、done=true，reason 为物体距参考或腕超过 0.25 m。评分字段 graded_by_live=1、sim_steps=35、task_failure=1，与此最后局一致。"
+     },
+     {
+      "t": 1.4,
+      "log_step": 212,
+      "text_zh": "推断：视频中未完成持兔运动，且最后评分局仅执行 35 个控制步便终止，未达到 DAG 要求的完整 99 步，因此本 event 节点从未完成；不依据 success=0 单独反推过程。控制步和视频播放时长未直接换算。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–1.40 s 可见 MyoHand 单手向兔子附近伸展、收指，但没有形成可确认的对向夹持或承载抓握。指尖与兔子局部轮廓重叠，两个视角也不能可靠区分短暂轻触与投影重叠；无接触对或力记录，故实际接触方式记未知。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.88,
+   "log_step": 205,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80–1.16 s，机器人手指在兔子附近收拢，但未将兔子夹持提起；随后手兔分离，直到视频 1.40 s 均未修复。log step 205 记录此局在控制步 35 因物体距参考或腕超过 0.25 m 而结束；log step 212 确认它是最后第 11 局。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.88,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–1.40 s 未见机器人建立能承载兔子的对向抓握；视频约 0.80–1.16 s 手指收拢后，兔子仍在桌面并倾斜，视频 1.40 s 手与兔子明显分离。这里未建立的是抓取所需的承载接触，不排除短暂指尖轻触。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 199 将手部位姿参考冻结，log step 204 在此基础上以固定伸手偏移及定时肌肉闭指执行 g1，没有以兔子是否被包夹来反馈修正抓取；视频约 0.80–1.40 s 显示收指未包住兔子，手兔随即分离。目标是抓取兔子，但实际伸手与闭指配合未建立抓握，支持控制层归因；现有证据不能分清接触几何设计与关节跟踪误差，不细分控制子类。"
+   }
+  ],
+  "gap_zh": "无状态日志和成功参考视频；评分字段及运行日志未报告 track_pos_err、track_rot_err、track_ok、track_max，故四项原生指标均置 null，无法复核逐步误差。已检查全部 36 帧及最后一秒；可排除成功承载抓握，但不能确定局部遮挡处是否有短暂指尖轻触，contact_actual 记未知。视频播放时间未与 50 Hz 控制时钟建立映射，所有视频锚点直接取自检查帧。最后评分局启用了冻结手部参考的诊断设置（log step 199、204）；log step 209 的后续测试在重置时即被拒绝（log step 210），不属于评分视频，终局后打包的脚本也不是另一次评分轨迹。",
+  "unmet": {
+   "bunny_motion_tracked": false
+  }
+ },
+ "summary_zh": "最后评分的第 11 局中，机器人尝试伸手闭指，但未形成承载抓握；兔子留在桌面并倾斜，唯一节点“兔子沿参考运动”未完成（0/1）。断点在抓取操作中，伸手与闭指控制未将兔子包夹，服务在控制步 35 提前判负；随后请求重置被拒绝，未改变评分局。缺少状态接触对、原生误差数值及成功参考，轻触方式和控制子类无法确定。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "唯一节点未完成、断点、接触未知和控制归因均与视频及 log step 212 的服务状态（control_steps=35, done）一致；未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:05:37Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--rtry8gj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--rtry8gj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "76dde31af55bf845dde59e2ad681997c9b88df3fadd42bcf056443441d8ecb73",
+   "video_sha256": "be54a7f4f4234261d598b0f2d5ba12ffc19635bb1c16fa54431e3086d50c1b8c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--shnioas.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--shnioas.json
new file mode 100644
index 0000000000000000000000000000000000000000..ab7f8e8d8e002817006560d1ff26d3b340c4c3db
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--shnioas.json
@@ -0,0 +1,240 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-standard--shnioas",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 41,
+   "sim_steps": 41,
+   "physics_steps": 410,
+   "steps": 389,
+   "calls": 417,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 389,
+   "elapsed_s": 2801.4,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.081,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.223,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-standard--shnioas/replay.mp4",
+  "duration_s": 1.68,
+  "fps": 25.0,
+  "frames": 42,
+  "sha256": "c2bae1f8e5b957e209fad8bae7e09731d7cc81c6343093b795a414ecc96b77d0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.56,
+      "log_step": 122,
+      "text_zh": "观察：视频 0.00–0.56 s，MyoHand 手指在兔子后侧闭合，兔子大体维持桌上的初始姿态；log step 122 的服务状态为 episode=11、control_steps=14、resets_left=0，确认这是最后一局。"
+     },
+     {
+      "t": 0.8,
+      "log_step": 125,
+      "text_zh": "观察：视频 0.72–1.20 s，手上移、弯曲的指端与兔子逐渐分离，兔子前倾翻转；视频 1.20–1.64 s，兔子继续远离手，未被重新夹住。log step 125 的服务终局为 episode=11、control_steps=41、done=True，reason 为物体距参考或手腕超过0.25米；评分字段 n_actions=41、live_success=0。"
+     },
+     {
+      "t": 1.64,
+      "log_step": 125,
+      "text_zh": "推断：最后一局只执行41个控制步并提前终止，未完成节点所需99步的完整参考跟踪，故此 event 节点始终未达到。视频锚点均取已检查帧的索引除以25 fps；视频长度与50 Hz控制步的仿真时长不同，未将控制步乘0.02直接当作视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.40–0.80 s，MyoHand 弯曲的手指从兔子后侧及下侧接触、拨动兔子，兔子随手移动后倾转，随后与上移的手指分离。推断：接触曾短暂带动物体，但没有持续保持承载；未能确认拇指与其余手指的对向夹持，实际可见接触仍记为推/扫。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.8,
+   "log_step": 125,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80 s，MyoHand 开始明显与兔子分离，兔子倾转而没有保持随手抬举；此后至视频 1.64 s 未恢复夹持。log step 125 的最后一局服务状态报告41个控制步后因物体距离超限而失败，未完成99步参考运动。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.40–0.72 s，MyoHand 弯曲的手指贴在兔子后侧及下侧，兔子随手移动；视频 0.72–1.04 s，手继续上移，兔子相对手指倾转并逐渐脱离，至视频 1.20–1.64 s 明显远离手，未被重新接住。推断：接触曾短暂带动物体，但未能在抬举和转向时保持承载，兔子滑转脱离，属于“没保持住”；局部遮挡不足以确认各指是否形成过对向夹持。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "log_step": 124,
+    "evidence_zh": "推断：用于抬举的接触布局不足以持续承载并约束兔子的转动。视频 0.40–0.88 s 可见手指从兔子后侧及下侧带动它，随后兔子相对手指倾转，未能保持随手抬举；log step 124 中 agent 的 contacts.py 估算末节拇指距兔子凸包约0.0149米，而食指、中指末节估算距离约为0。这是 agent 根据相机拟合和手模型计算的几何估计，并非模拟器接触日志；它与可见的偏侧接触及随后滑转一致，支持接触设计归因，但未验证可行替代握法。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "没有状态日志和成功示范视频；兔子与手的局部接触受遮挡，无法精确确认各指接触对、接触力或是否存在极短暂夹持。log step 124 的接触距离与 log step 124–125 的 CAMERA ERROR 均来自 agent 自行拟合，不能作为评分器原始指标。run.json 的 reward 未报告 track_pos_err、track_rot_err、track_ok / track_steps 或 track_max，故四项原生指标均为 null；完整跟踪未完成则由可见分离和最后一局服务提前终止共同确认。"
+ },
+ "summary_zh": "最后评分的第11局中，MyoHand 的手指接触并短暂带动兔子，但未保持住；视频约 0.80 s 后兔子翻转并与手分离，最终完成0/1节点。服务在最后一局第41个控制步因物体距离超限终止，未完成99步参考跟踪，接触布局不适配是有视觉与日志估计支持的根因。缺少状态接触记录及评分器原始跟踪误差，无法核实各指夹持关系、精确接触力与容差数值。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "断点、节点值、时钟均无误；问题在现象：兔子在视频 0.4–0.9 s 被手带着走并倾转（有接触、被带动后丢失），应为没保持住而非没建立接触；根因 抓取和接触设计 可保留。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达到、断点、时钟均与视频和 log step 122/125 的服务读数一致；现象已改为没保持住，根因（抓取和接触设计）有视频与 step 124 估计支持且标明为推断，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:13:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--shnioas/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--shnioas/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--shnioas/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--shnioas/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1509a043a24ec744a76640a4f2760307a9f6b1c8c94b9ab25c581855867bed5c",
+   "video_sha256": "c2bae1f8e5b957e209fad8bae7e09731d7cc81c6343093b795a414ecc96b77d0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--vuotozw.json b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--vuotozw.json
new file mode 100644
index 0000000000000000000000000000000000000000..2d895bf1dde2fca452253ab03aa1babfe23726a2
--- /dev/null
+++ b/phase2/myosuite/dex_bunny_inspect/trajectories/myosuite-dex-bunny-inspect-i00-standard--vuotozw.json
@@ -0,0 +1,211 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-bunny-inspect-i00-standard--vuotozw",
+ "taskKey": "myosuite/dex_bunny_inspect",
+ "dag_sha256": "a0345a5ca5780c5190affd3ffde1168212a7a9991e0d4ba187c105b66032d4c4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 9,
+   "sim_steps": 9,
+   "physics_steps": 90,
+   "steps": 78,
+   "calls": 102,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 78,
+   "elapsed_s": 1251.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.464,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.586,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-bunny-inspect-i00-standard--vuotozw/replay.mp4",
+  "duration_s": 0.4,
+  "fps": 25.0,
+  "frames": 10,
+  "sha256": "9bfda7a2c19e47f53a3d1792fc15e5ded284aeb4ac2d71933918313e637d1448"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bunny_motion_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.36 s 的全部10帧及两个并列视角，兔子一直留在桌面；MyoHand 起初在兔子后上方屈指，随后空手退离，未见兔子随手抬起并连续转向。"
+     },
+     {
+      "t": 0.36,
+      "log_step": 165,
+      "text_zh": "观察：log step 165 中 gen.py 将第10局重置为第11局并提交99步动作，机器人服务仅执行9个控制步即返回 done=True，原因为物体距离参考或手腕超过0.25米；reward 的 episodes=11、n_actions=9、graded_by_live=1 与该局一致。该局未完整执行节点要求的99步，视频末帧亦显示手与桌上兔子分离。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–0.08 s 可见单只 MyoHand 的手指在兔子后上方屈曲，之后退离；兔子未被包住并带离桌面。起始手指与兔子背部在投影中重叠，两个视角均不能确认是否短暂触碰及其接触方式；不据典型方案认定实际抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok / track_steps",
+   "unit": "fraction",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bunny_motion_tracked",
+   "t": 0.12,
+   "log_step": 165,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.08–0.20 s 中，MyoHand 手指继续屈曲而手掌逐渐离开兔子，兔子仍在桌面；至视频 0.36 s 手与兔子明显分离，未恢复抓持或开始抬起。log step 165 的机器人服务报告第11局在9个控制步后因物体距离参考或手腕超过0.25米而终止，跟踪步骤未完成。"
+  },
+  "unmet": {
+   "bunny_motion_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.12,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.36 s 的两个视角中，MyoHand 在兔子后上方屈指后空手退开，兔子始终由桌面承托，未见手指形成能将兔子带起的对向夹持。推断：本步骤所需的有效抓持未建立；起始处可能的短暂擦碰不足以证明曾持有后滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 163 的实际执行脚本将预计算关节轨迹相对起点的差值作为六个根部执行器目标，并配合固定屈肌、腕肌和旋前肌输入，一次性提交动作，无执行中的位置或抓持反馈修正；log step 166 检查实际加载的轨迹，前10行根部差值全为零，仅脚本前6步另加逐渐衰减的前移、下压偏置。结合视频 0.00–0.36 s 手空收并退离桌上兔子，以及 log step 165 的提前终止，可归因于该控制实现未将手保持在可建立夹持的位置；现有证据不足以进一步区分接触几何与动态跟踪问题。"
+   }
+  ],
+  "gap_zh": "无状态日志或成功示范；起始手指与兔子背部局部遮挡，无法确定短暂实际接触的方式，故 contact_actual 保留未知。reward 未报告四项原生跟踪指标，均填 null；服务终止提示未区分距参考过远与距手腕过远。日志9个20毫秒控制步与视频10帧、0.40 s 的时长不直接相等，未将控制步换算为视频时刻，断点仅采用已检查画面的粗略锚点。"
+ },
+ "summary_zh": "评分的第11局未完成唯一节点“兔子沿参考运动”：视频全段显示兔子留在桌面，MyoHand 屈指后空手退离，机器人服务在9个控制步后提前终止。失败发生在建立有效抓持并开始跟踪的操作中，日志中的整批固定输入与画面中的手部退离支持控制层面的归因。无状态日志和成功示范，初始短暂接触方式、精确跟踪指标及控制子类无法确定。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "视频仅10帧（0.4 s），兔子留在桌面、手屈指后退离，与标注一致；节点未达成、断点与现象/根因有证据支持。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:05:15Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--vuotozw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-bunny-inspect-i00-standard--vuotozw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "effc53d1a0ef9685849c3453beef5d1cce473ac64674ca33f9147c9ef656324e",
+   "video_sha256": "9bfda7a2c19e47f53a3d1792fc15e5ded284aeb4ac2d71933918313e637d1448",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--7xl4jsm.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--7xl4jsm.json
new file mode 100644
index 0000000000000000000000000000000000000000..b126fc0687839942f3478cff5fef81e1069a8d37
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--7xl4jsm.json
@@ -0,0 +1,204 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-privileged--7xl4jsm",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 81,
+   "sim_steps": 32,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 77,
+   "agent_resets": 18429,
+   "agent_sim_steps": 975,
+   "agent_records": 0,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.85,
+   "peak_processes": 30,
+   "peak_disk_gb": 0.248,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-privileged--7xl4jsm/replay.mp4",
+  "duration_s": 1.32,
+  "fps": 25.0,
+  "frames": 33,
+  "sha256": "4835f6b9d9aca7e2c016896851d88daf72050364833cc98f9317d591e2aa2030"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.80 s，MyoHand 的手指逐渐合拢围住粉色杯子并带动杯子小幅移动；没有完成提起、倾斜倒杯再回落的整段运动。"
+     },
+     {
+      "t": 1.04,
+      "log_step": null,
+      "text_zh": "观察：视频 0.92–1.04 s，杯子与手指间出现并扩大间隙；视频 1.16–1.28 s，杯子在手下方独自倾斜下落，手未重新抓住杯子。"
+     },
+     {
+      "t": 1.28,
+      "log_step": 87,
+      "text_zh": "观察：log step 87 调用最终参数的保存脚本，输出原生失败原因“object more than 0.25 m from the reference (or from the wrist)”，实际生成的运动为32个控制步；inputs/run.json 的 reward.sim_steps=32、n_actions=81、task_failure=1、replay_success=0、deterministic=1。评分回放提前结束，未完整执行节点所需的81个控制步；视频末帧亦未完成倒杯运动。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.24–0.80 s，MyoHand 单手的拇指与其余弯曲手指从相对两侧围夹粉色杯子的杯壁及杯口附近；视频约 0.92–1.04 s 抓握失去，随后手杯分离。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.92,
+   "log_step": 87,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.80 s 杯子仍处于手指围夹区域，视频 0.92–1.04 s 杯子逐渐脱离手指，随后在视频 1.16–1.28 s 倾斜下落，未恢复抓握或完成整段参考动作。log step 87 的最终保存与回放输出原生失败；评分器记录实际只执行32个控制步。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 0.92,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.24–0.80 s，MyoHand 的拇指和弯曲手指围夹杯子；视频 0.92–1.04 s 杯子从手指围夹区脱离，视频 1.16–1.28 s 杯子与手完全分开并向下倾倒。机器人没有在倒杯运动中持续保持杯子，也没有重新抓住。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 65、68 的 agent 控制脚本把前臂参考运动转为位置伺服前馈/反馈命令，并用肌群激活与手部力矩拟合控制抓握；log step 69 的保存脚本记录这些动作并重新回放，log step 87 用最终参数执行该脚本仍得到原生失败。结合视频 0.92–1.28 s 的手杯分离与掉杯，所交付的前臂和手指动作未维持搬动杯子所需的抓握，支持控制问题；现有证据不足以细分为接触设计、反馈控制或参数问题。"
+   }
+  ],
+  "gap_zh": "未提供状态日志及成功参考视频，无法核实各手指的接触力、精确脱离时刻，以及握法与前臂跟踪误差各自的贡献。评分器 reward 未报告 track_pos_err、track_rot_err、track_ok、track_max，四项 native_metrics 均留空；log step 87 的最终保存试跑另输出 track_ok=16，但没有最终评分回放的完整误差序列。log step 89 中 agent 关于执行器能力不足的解释未排除所有合法控制方案，不能据此判为任务或环境问题。所有时间锚点均直接取自视频，没有把20毫秒控制周期或 agent 工作时间换算为视频时间。"
+ },
+ "summary_zh": "MyoHand 围夹杯子并尝试带动它，但在视频约 0.92–1.04 s 失去抓握，杯子随后倾斜下落，唯一节点“完成倒杯轨迹”未达成（0/1）。最终保存记录及评分器确认只执行32/81个控制步便原生失败，根因保守归为控制。缺少状态日志和最终评分的精确追踪指标，无法进一步区分接触设计与前臂跟踪问题。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频、log 一致：节点未达成，杯子在视频约0.9–1.0 s 脱离手指后掉落，断点、没保持住与保守的控制根因均有证据支持，未发现时钟换算问题。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:37:45Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--7xl4jsm/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--7xl4jsm/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6d2b178b188c3d318f284fd899dc733b8f527dda404515244215862b38e7b123",
+   "video_sha256": "4835f6b9d9aca7e2c016896851d88daf72050364833cc98f9317d591e2aa2030",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--bc66v9p.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--bc66v9p.json
new file mode 100644
index 0000000000000000000000000000000000000000..6e8c8bf63fffb437f3f234cd460c7a582614d217
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--bc66v9p.json
@@ -0,0 +1,204 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-privileged--bc66v9p",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 81,
+   "sim_steps": 58,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 70,
+   "agent_resets": 21,
+   "agent_sim_steps": 802,
+   "agent_records": 0,
+   "agent_replays": 20,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.708,
+   "peak_processes": 19,
+   "peak_disk_gb": 0.84,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-privileged--bc66v9p/replay.mp4",
+  "duration_s": 2.36,
+  "fps": 25.0,
+  "frames": 59,
+  "sha256": "ba3e8894d96788f8a45788059ef902507cb8ba8c0a5ac8b597c68578bfaa1fcb"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.4,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频0.00–2.32 s的全部59帧。视频0.00–0.76 s，MyoHand单手贴近粉色杯子并带动其离开原位、逐渐倾斜；视频0.96–1.12 s手与杯子短暂分开，视频1.28–1.84 s手指再次贴近杯壁。"
+     },
+     {
+      "t": 2.12,
+      "log_step": null,
+      "text_zh": "观察：视频1.84–2.00 s手指仍贴着倾斜的杯子；视频2.08–2.32 s杯子与手分离、翻转下落，空手停在其上方，直到末帧未重新控制杯子，也未完成回落轨迹。"
+     },
+     {
+      "t": null,
+      "log_step": 171,
+      "text_zh": "观察：log step 166的收尾脚本选择并保存81行动作，log step 169的FINAL回放状态为success=False、done=True、steps=58、track_ok=39，原因是杯子距参考位置或手腕超过0.25米；log step 171读取交付文件的两次独立回放结果，两次均在第58个控制步终止且终态哈希相同。run.json的reward也给出sim_steps=58、replay_success=0、deterministic=1，与视频中杯子脱离后的提前结束一致；该节点要求完整执行81步，因此未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频0.40–0.76 s及视频1.28–2.00 s，MyoHand手指贴压杯壁一侧，带动杯子移动和倾斜；视频2.08–2.32 s接触丢失。画面可辨认侧向手指接触，但不能确认拇指与其余手指持续形成对向夹持，不把这种贴压记为稳定抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 2.08,
+   "log_step": 169,
+   "where": "操作中",
+   "evidence_zh": "观察：MyoHand正在执行倒杯轨迹的回落运动；视频1.84–2.00 s手指尚贴着杯子，视频2.08–2.32 s杯子脱离、翻转下落，手未重新留住它。log step 169记录最终选中动作的回放在第58个控制步原生失败终止，log step 171的两次交付文件回放复现该结果，未达到完整81步的验收条件。视频锚点来自画面，不由agent工作时间换算。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.08,
+    "log_step": null,
+    "evidence_zh": "观察：视频1.84–2.00 s杯子仍与MyoHand手指贴靠并随手回落；视频2.08–2.16 s杯子从手指下方脱离，视频2.20–2.32 s继续独自翻转下落，手中无杯，未恢复接触。这里是搬运过程丢失杯子，不是放到目标后按计划松手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：视频1.84–2.32 s显示MyoHand在回落和转动时未能维持对杯子的控制。log step 161的候选生成脚本在后段沿用手指执行器指令并调整前臂，log step 169记录最终选中该候选的终态哈希及第58个控制步失败，log step 171的两次独立回放得到相同终态；结合agent在log step 153自述单侧手指支承而拇指未贴合、log step 166自述完整回放仍失败，支持手指与前臂动作未稳定留住杯子的控制问题。现有证据不足以进一步区分抓取接触设计、轨迹求解或反馈控制子类。"
+   }
+  ],
+  "gap_zh": "无状态日志或成功参考视频，不能核验各手指的实际接触对、接触力及逐步未舍入轨迹误差，控制根因仅标到大类。run.json的reward没有报告DAG要求的四项原生指标，native_metrics相应填null；log step 169中交付候选回放显示的末步位置误差0.27288米、旋转误差0.865弧度和track_ok=39仅作回放证据，不补作评分器reward字段。视频按25 fps直接取时；未将20毫秒控制周期或agent工作时间直接映射到视频。"
+ },
+ "summary_zh": "机器人带动杯子移动和倾斜，但回落过程中未能持续留住杯子，视频约2.08 s后杯子脱离并下落，完成倒杯轨迹节点最终为0/1。交付的81行动作在两次独立回放中均于第58个控制步提前失败，根因保守归为控制。缺少状态日志、成功对照及完整精确误差，无法确认对向抓握或细分控制原因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "单节点未达成、断点在cup_pour_tracked、没保持住加控制根因均与视频（约2.0–2.1 s杯子脱手下落）及log step 169/171的第58步失败一致；时间均为视频时间，无换算问题，无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:38:29Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--bc66v9p/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--bc66v9p/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "53a2c5fa26d1858bb47beb3543cfa03dea2159c6ea4803ce22bb5a4338d4d1c8",
+   "video_sha256": "ba3e8894d96788f8a45788059ef902507cb8ba8c0a5ac8b597c68578bfaa1fcb",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--bx8vp9q.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--bx8vp9q.json
new file mode 100644
index 0000000000000000000000000000000000000000..e567a407769c5491e95afe7255b780af84d24820
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--bx8vp9q.json
@@ -0,0 +1,197 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-privileged--bx8vp9q",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 22,
+   "sim_steps": 22,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 108,
+   "agent_resets": 210,
+   "agent_sim_steps": 3326,
+   "agent_records": 139,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.751,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.245,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-privileged--bx8vp9q/replay.mp4",
+  "duration_s": 0.92,
+  "fps": 25.0,
+  "frames": 23,
+  "sha256": "3f4eb4ada6ff6fdb849e365ac7ba9cb59f62d5e5a3feb84f6af677d5f96732bd"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.52,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频0.00–0.88 s的全部23帧；MyoHand手指围在粉色杯子的杯壁两侧，杯子只在桌面附近小幅移位、倾斜，视频0.52–0.88 s未出现随手抬高后完成倒杯和回落的运动。"
+     },
+     {
+      "t": 0.88,
+      "log_step": 132,
+      "text_zh": "观察：log step 87的生成脚本保存22×45动作；log step 132对该交付文件的两次回放均在控制步22终止，原因均为杯子距参考或手腕超过0.25米，终态哈希相同；inputs/run.json的评分字段也报告n_actions=22、sim_steps=22、task_failure=1。推断：结合视频未出现完整倒杯运动，本次没有完成要求的81步轨迹。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频0.12–0.88 s，单只MyoHand的拇指和其余手指从杯壁两侧包围合拢，杯子随接触小幅移位；未见杯子被稳定提离桌面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.52,
+   "log_step": 132,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约0.52 s起至0.88 s，MyoHand仍围着杯子运动，杯子留在桌面附近，仅小幅移位、倾斜，未进入明显抬杯倒杯运动；该问题直到视频结束未恢复。log step 132对交付轨迹的重复回放报告控制步22提前失败，评分字段确认只执行22步。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.52,
+    "log_step": null,
+    "evidence_zh": "观察：视频0.00–0.88 s，MyoHand接触杯壁并使粉色杯子小幅移位、倾斜，但视频0.52–0.88 s杯子仍停留在桌面附近，没有随手形成任务要求的抬杯、倒杯轨迹。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "观察：log step 87保存的控制器用参考手关节目标、前臂计算力矩控制和肌肉非负最小二乘控制生成动作；log step 132确认该文件重复回放均失败。视频0.12–0.88 s可见手指包围杯壁并运动，杯子却只在桌面附近小幅移位。推断：该控制器没有把手和前臂的动作转化为杯子所需的提起与轨迹运动；现有证据不足以区分接触承载、前臂跟踪或二者共同作用，根因保留在控制大类。"
+   }
+  ],
+  "gap_zh": "观察：inputs/中有视频及完整运行日志，但没有状态日志或成功示范视频；评分结果没有报告track_pos_err、track_rot_err、track_ok、track_max的原生数值。视频中的手部为半透明肌骨渲染，无法精确判定各手指接触力及抓握承载状态；日志后续诊断多为另外重置的会话试验，不用其接触读数替代评分轨迹。推断：只能确定未完成完整轨迹，不能可靠细分控制根因或量化逐步误差；断点视频0.52 s是可视失败的粗略导航锚点，未将控制步或agent工作时间换算成视频时间。"
+ },
+ "summary_zh": "观察：评分视频0.00–0.88 s中，MyoHand围住粉色杯子并使其小幅移位、倾斜，未完成抬起、倒杯和回落，唯一节点未达；交付文件的两次回放均在控制步22提前失败（log step 132）。推断：断点为完成倒杯轨迹的操作中，现象是位姿不准，根因归于控制；缺少评分轨迹的逐步状态、接触力和原生误差数值，不能进一步细分。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注与证据一致：视频0.00–0.88 s杯子停留在桌面附近，仅小幅移位、倾斜，唯一节点未达；log step 87保存22步轨迹，step 132两次回放均在控制步22失败，评分字段n_actions=22、task_failure=1。断点、位姿不准现象、控制根因及gap_zh均有依据，未发现需改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:38:49Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--bx8vp9q/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--bx8vp9q/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "be1cb8fd8cdf045b445974762cfadd7fbd59eb33940c1b109630df65ffc9efd7",
+   "video_sha256": "3f4eb4ada6ff6fdb849e365ac7ba9cb59f62d5e5a3feb84f6af677d5f96732bd",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--dbzqvvb.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--dbzqvvb.json
new file mode 100644
index 0000000000000000000000000000000000000000..6ed2ef47f0b3b34ff0b0c180b2d14ef00dfb2323
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--dbzqvvb.json
@@ -0,0 +1,204 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-privileged--dbzqvvb",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 81,
+   "sim_steps": 22,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 1290,
+   "agent_resets": 1316,
+   "agent_sim_steps": 14877,
+   "agent_records": 0,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 22.628,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.84,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-privileged--dbzqvvb/replay.mp4",
+  "duration_s": 0.92,
+  "fps": 25.0,
+  "frames": 23,
+  "sha256": "b389ac3899690fe11328715bb67deda16e45e79d084108ef0c269c6ee1eeeb16"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.32,
+      "log_step": null,
+      "text_zh": "观察：视频 0.16–0.64 s 的两个视角中，MyoHand单手弯曲手指合围粉色杯身，杯子贴着手指发生倾斜；这表明已开始操作杯子。"
+     },
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频 0.00–0.88 s 的全部23帧；视频 0.56–0.72 s，绿色位置标记与粉色杯身逐渐错开，末段标记遮住部分杯口，视频结束前未出现完整倒杯后回落过程。"
+     },
+     {
+      "t": null,
+      "log_step": 190,
+      "text_zh": "观察：log step 190 重新生成最终交付文件后进行两次独立回放，两次均为 success=false、steps=22、replayed_rows=22、rows=81，原因均为物体距参考位置或手腕超过0.25米，instance_ok=true、deterministic=true；reward也报告 sim_steps=22、task_failure=1。完整81步的节点验收未达。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.32–0.64 s，MyoHand单手弯曲手指从杯身两侧合围，带动粉色杯子倾斜；末段绿色标记遮挡部分接触区域。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.6,
+   "log_step": 190,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.60 s，机器人已接触并倾斜杯子，但粉色杯子与绿色位置标记明显错开，此后没有完成整段倒杯运动。log step 190 的最终交付回放在控制步22因物体距参考位置或手腕超过0.25米提前结束；断点属于正在执行的倒杯轨迹步骤，视频锚点取自画面，未由控制步时长换算。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.6,
+    "log_step": 190,
+    "evidence_zh": "观察：视频 0.48–0.72 s，粉色杯子仍贴近手指作局部倾斜，绿色位置标记却逐渐移到杯外；视频 0.80–0.88 s 标记进一步移向镜头并遮挡杯口。log step 15 显示环境将参考物体位置写入目标site，log step 190 的最终回放报告物体距参考或手腕超限；推断：机器人带杯运动没有跟上参考位姿。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 90 的会话试验脚本按参考位移更新前臂目标，并加入位置/速度误差反馈；但 log step 93 的 k=2 试验中，试验控制步12–22的腕部旋转q3约停在-0.40至-0.39弧度，杯高约0.067–0.069米，而参考高度升至0.286米。该试验末态哈希 b9f06207b479fe1f 与 log step 190 最终文件两次回放的末态相同，视频 0.48–0.72 s 也显示杯子运动不足以跟随参考，支持带杯控制未能实现所要求的运动；现有证据不足以细分为握法、轨迹求解或反馈参数问题。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志、接触对或成功示范；视频末段绿色标记遮挡杯口及部分手指，不能据此确定末态是否仍稳定夹持，也不能断言脱手。reward未报告track_pos_err、track_rot_err、track_ok、track_max，四项native_metrics均保留null；log step 93的同末态会话试验只用于支持控制归因，不替代正式评分指标。未记录最终交付的完整目标、执行器及接触力序列，控制子类不能进一步确定。"
+ },
+ "summary_zh": "观察：MyoHand手指合围杯身并带其倾斜，但视频中杯子与绿色位置标记逐渐错开，完整倒杯轨迹节点始终未达，终态为0/1。观察：log step 190 的81行动作文件两次回放均只执行22步即触发原生距离失败；推断：结合 log step 90、93 中参考驱动命令与实际腕部运动不符的试验记录，断点属于倒杯轨迹操作中的控制问题。缺少逐帧状态、完整接触力和正式原生指标，无法确定更细控制根因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与证据一致：视频仅23帧（0.92 s），绿色参考标记逐渐离开几乎未升起的粉色杯子；log step 190 两次回放均在22步因距离超限终止，step 93 试验末态哈希相同，杯高约0.069 m而参考升至0.286 m。节点未达、断点、位姿不准与控制归因均有依据，推断已标明。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:38:44Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--dbzqvvb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--dbzqvvb/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3152d8fac7b482ff1a70095b8c1aa50fd9f260cefec6e650457fce9db23f2bf6",
+   "video_sha256": "b389ac3899690fe11328715bb67deda16e45e79d084108ef0c269c6ee1eeeb16",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--ysubn6s.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--ysubn6s.json
new file mode 100644
index 0000000000000000000000000000000000000000..743ab8fff31efaa4133c28d844fa2e89f7d20f5f
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--ysubn6s.json
@@ -0,0 +1,234 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-privileged--ysubn6s",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 22,
+   "sim_steps": 22,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 64,
+   "agent_resets": 183,
+   "agent_sim_steps": 588,
+   "agent_records": 0,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.957,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.984,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-privileged--ysubn6s/replay.mp4",
+  "duration_s": 0.92,
+  "fps": 25.0,
+  "frames": 23,
+  "sha256": "1a0645063f941385670ad68da699057cd43e7aab916e1ff29ababb1e22587aec"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的两个外部视角显示粉色杯子立在桌面，MyoHand手指位于杯壁旁；这只是预抓取初态，尚未完成整段倒杯运动。"
+     },
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：已按顺序检查视频 0.00–0.88 s 的全部23帧；MyoHand手指在杯壁旁屈曲，杯子有小幅平移和晃动，但始终停在桌面附近，没有出现随手抬起、明显倒杯及回落的完整运动；视频 0.60、0.72、0.80、0.88 s 均仍见杯子留在原工作区。"
+     },
+     {
+      "t": null,
+      "log_step": 151,
+      "text_zh": "观察：log step 151 的最终生成脚本输出实际执行22个控制步、track_ok=14，随后对同一动作数组的Session.replay仍在22步失败并报告track_ok=14，保存数组形状为(22,45)。track_ok是任务服务对位移和旋转同时达标的控制步计数，低于DAG要求的73步，也未执行完81步。"
+     },
+     {
+      "t": null,
+      "log_step": 153,
+      "text_zh": "观察：log step 153 对最终保存的trajectory.npz进行两次正式评分器回放，两次均报告success=false、steps=22、同一终态哈希fbe9288eff21e6fe，原因为物体距参考位置或手腕超过0.25米；run.json的reward同时报告sim_steps=22、deterministic=1、instance_ok=1。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.20–0.88 s，MyoHand单只手的手指沿杯壁旁屈曲，伴随杯子小幅移动和晃动；实际可见的是手指对杯侧的推拨，未见明确的对向夹紧和承载抬杯。半透明手模型使具体指节接触点难以分清。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": 14
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.6,
+   "log_step": 151,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.60 s 起，杯子仍在桌面附近，之后直到视频 0.88 s 都未随手抬起完成倒杯；log step 151 的最终交付只执行22步，log step 153 的正式回放也在22步因物体距离超限终止。推断：失败发生在正在进行的倒杯轨迹跟随步骤，前期的手指接触没有转化为所需的杯子运动。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.6,
+    "log_step": 153,
+    "evidence_zh": "观察：视频 0.20–0.88 s 的两视角中，MyoHand手指在粉色杯侧屈曲，杯子仅小幅移动、晃动；视频 0.60、0.72、0.80、0.88 s 均未见杯子被包握带离桌面。推断：未建立的是倒杯运动所需的承载接触，已有侧向推拨不等于成功抓握，亦无先承载后滑脱的证据。观察：log step 153 的两次正式回放均在22个控制步因物体距参考位置或手腕超过0.25米终止；此字段确认提前失败，本身不证明具体接触故障。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "t": null,
+    "log_step": 151,
+    "source_zh": "日志",
+    "evidence_zh": "推断：预抓取位的手指合拢方案未形成能承载杯子的包握接触。观察：视频 0.20–0.88 s，手指在杯侧屈曲并伴随杯子小幅移动，杯子没有被手带起。log step 151 的交付生成代码将cand2.npy作为wrap_fingers.npy的抓握目标；同一目标在log step 145的会话诊断中已实际尝试抬臂，控制计数k=50时腕部lunate高度为0.143米、杯子高度为0.053米，杯子全程最大高度仅0.058米，显示抬臂仍未带起杯子。该诊断是会话中的辅助根因证据，不是评分轨迹的状态。log step 151及153确认交付在22个控制步提前失败。推断：结合画面中的侧向推拨与相同抓握目标的诊断，支持抓取和接触设计问题；不能据此确定具体指节接触力或证明所有抓法不可行。",
+    "alt_feasible": "未知"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功示范；两个视角的半透明手模型不能精确确认各指节接触点及接触力。抓取和接触设计为视频与同一抓握目标会话诊断支持的推断，尚不能区分具体几何、肌肉跟踪及承载力问题，也未验证可行替代方案。log step 151 的抬升分支确实要到控制计数k>30才启动，而交付在22步终止；未单独验证提前抬升的效果，不能排除时序另有影响，不再将其列为主根因。最终回放的track_ok=14见log step 151；评分结果未报告交付回放的track_pos_err、track_rot_err和track_max数值，三项指标留null。视频共0.92秒，而22个20毫秒控制步对应仿真0.44秒，未建立两者精确映射；所标视频时刻均来自已检查帧，不由控制计数换算。"
+ },
+ "summary_zh": "评分轨迹中MyoHand手指在杯侧屈曲并推拨杯子，未建立带杯抬升的承载接触，完成倒杯轨迹节点未达成，终态为0/1。log step 151及153确认交付在22个控制步因物体距离超限失败；结合视频与log step 145对同一抓握目标的抬臂诊断，推断根因为控制·抓取和接触设计。置信度由8降至7：缺少状态接触数据，具体承载失效机制未确定，亦未排除抬升时序的影响。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达成、终态0/1、断点 cup_pour_tracked 均正确，时间与 log step 存在；但根因「推理/规划（抬升分支与参考不同步）」只是对生成脚本时序的推断，视频和交付日志指向的是抓取没有建立：手指合拢时把杯子推开，杯子从未被手承载。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、终态0/1、断点cup_pour_tracked、没建立接触现象与抓取和接触设计根因（标为推断）均有视频和log step 151/153支持；时钟均标明视频时间，未转换agent时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:09:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--ysubn6s/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--ysubn6s/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--ysubn6s/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--ysubn6s/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "bb5e0fa95b7f404424f4274b799ea9fc0a18ba96b0005bf729ec45de292f5bc0",
+   "video_sha256": "1a0645063f941385670ad68da699057cd43e7aab916e1ff29ababb1e22587aec",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--zfhuyuj.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--zfhuyuj.json
new file mode 100644
index 0000000000000000000000000000000000000000..be88f05abb9d78bf5bf7f7c5f54fd1feb8ca8c78
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-privileged--zfhuyuj.json
@@ -0,0 +1,206 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-privileged--zfhuyuj",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 19,
+   "agent_resets": 65,
+   "agent_sim_steps": 1109,
+   "agent_records": 0,
+   "agent_replays": 7,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.644,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.239,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-privileged--zfhuyuj/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "05128412eb268c51c854eff37351caded433c591bd6e809ffc35101c7209dd33"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.28,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.76 s 的全部20帧及两侧视图，MyoHand手指起初贴近杯侧并伸展，杯子向远离手掌的方向移动；视频 0.28 s 起手与杯之间出现间隙，之后间隙扩大。杯子没有被合手夹住并随手完成离桌、倾倒、回落。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 53,
+      "text_zh": "观察：视频 0.76 s 末帧中，杯子仍在桌面附近，与张开的手指分离；log step 53 中agent复制最终轨迹文件后，评分器CLI回放该文件报告 n_actions=19、steps=19、success=false、instance_ok=true、deterministic=true，并以杯子距参考位置或手腕超过0.25米为失败说明；run.reward同样记录19步及task_failure=1。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 53,
+      "text_zh": "推断：交付回放只执行19个控制步，且画面缺少杯子的完整提起倾倒运动，因此未达到完整81步倒杯轨迹节点；不把初始贴近参考位置或会话测试中的局部达标计作该事件完成。视频时间直接取帧时间，未将log中的控制步或agent工作时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–0.28 s，MyoHand单手的手指在杯子靠手的一侧接触、拨动杯子，随后手指与杯子分离；未见拇指与其余手指形成包围或对夹，故实际接触记为推/扫。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.32,
+   "log_step": 53,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.28–0.40 s，MyoHand手指已经离开杯侧，杯子未被手带起，此后至视频 0.76 s 未重新抓住；log step 53 的最终文件回放报告19步即原生失败终止，不能完成81步倒杯追踪。断点锚定手杯明显分离且未形成抓握的时段。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.32,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.76 s，MyoHand手指只在杯子一侧拨动，随后张开并与杯分离，没有形成使杯子随手离桌的包围或对向夹持。推断：缺失的是任务所需的有效抓握接触；不是已抓起后又脱手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 0.32,
+    "log_step": 52,
+    "evidence_zh": "推断：agent脚本在log step 35、45、52中以参考关节轨迹、手指PD到肌肉有界最小二乘分配及前臂迭代学习生成动作；log step 52 的会话测试在符号修正后仍报告手部跟踪最大误差约0.67弧度、前臂约0.17，最终选出的回放仍为19步失败，log step 53确认复制后的交付文件也19步失败。视频 0.28–0.76 s 显示手指离杯而未完成抓握，支持该控制方案未实现抓持和搬动。会话诊断仅用于归因，不能据此认定本次逐帧关节误差，也不足以细分接触设计、轨迹或反馈控制。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "无状态日志或成功示范；透明手部模型及低分辨率画面不能确定具体接触指节、接触力和肌肉激活。会话诊断包含多次重置的其他测试，不能把它们的误差时序或接触数当作交付轨迹的状态；无法进一步区分控制子类。run.reward未提供DAG四项原生指标，故native_metrics的值保留null；log step 52虽报告最终回放track_ok=14，但缺少最终交付文件的完整未舍入逐步误差。视频时长与19个20毫秒控制步的仿真时长不相等，未建立跨时钟映射。"
+ },
+ "summary_zh": "评分轨迹中，MyoHand手指在杯侧拨动后与杯子分离，杯子未随手完成提起、倾倒和回落，完成倒杯轨迹节点最终为0/1。断点在倒杯操作中，现象为未建立有效抓握，日志与画面支持控制层面的失败；最终交付文件经确定性回放仅执行19步即触发原生失败。缺少状态日志、成功示范及完整精确误差，无法细分控制根因或核验全部原生指标。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、没建立接触与控制根因均有视频和log支持；视频仅0.8 s（19步），时钟未换算，校验通过。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:36:16Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--zfhuyuj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-privileged--zfhuyuj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6e535ff1eacd7b83856d460ebbf55824a619795be6745ad3dc83b24d026b0668",
+   "video_sha256": "05128412eb268c51c854eff37351caded433c591bd6e809ffc35101c7209dd33",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--ayq8nmx.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--ayq8nmx.json
new file mode 100644
index 0000000000000000000000000000000000000000..b4398cb291acec9eba3307fa62e37e6f05319d84
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--ayq8nmx.json
@@ -0,0 +1,211 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-standard--ayq8nmx",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 14,
+   "sim_steps": 14,
+   "physics_steps": 140,
+   "steps": 113,
+   "calls": 55,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 113,
+   "elapsed_s": 1076.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.563,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.841,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-standard--ayq8nmx/replay.mp4",
+  "duration_s": 0.6,
+  "fps": 25.0,
+  "frames": 15,
+  "sha256": "7c85ad9d9994f173f163c9ff5a3fcdaabcfb0bffbf68d8b730ca44325a54b626"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.2,
+      "log_step": null,
+      "text_zh": "观察：已检查视频 0.00–0.56 s 的全部15帧及手杯局部；MyoHand手指在杯身旁卷曲，杯子约从视频 0.16–0.20 s 开始明显倾斜，随后向远离手指的一侧滚动，视频 0.40–0.56 s 已与手部分离并滚向画面下缘，未见手持杯完成提起、倒杯和回落。"
+     },
+     {
+      "t": 0.56,
+      "log_step": 60,
+      "text_zh": "观察：log step 60 记录最后一次重置后的 episode=11，服务在 control_steps=14 时返回 done=true，原因为杯子距参考位置或手腕超过0.25米；inputs/run.json 的评分器字段为 graded_by_live=1、n_actions=14、success=0。该末回合的早停与视频末段手杯分离一致，未执行完节点要求的81个控制步；日志中的计划行号16不是实际执行步数。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.12–0.28 s，MyoHand单手的手指在杯身一侧卷曲并碰动杯子，杯子倾倒后向外滚离；未见手指从相对两侧夹紧杯子并将其稳定带起，按实际推碰记录。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.2,
+   "log_step": 60,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.20 s，MyoHand正在杯旁卷曲手指，杯子开始倾倒；视频 0.28–0.56 s 杯子持续滚离而手指留在后方，未恢复持杯。log step 60 的末回合 episode=11 在实际控制步14因杯子距参考或手腕过远而终止，因此断点为完成倒杯轨迹这一步，视频锚点取可见倾倒时刻。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "被挡/碰撞",
+    "t": 0.2,
+    "log_step": 60,
+    "evidence_zh": "观察：视频 0.12–0.28 s，MyoHand手指卷曲时碰动杯身，杯子向手外倾倒；视频 0.36–0.56 s 杯子继续侧翻、滚远，手指没有随杯保持包夹。这里记录的是操作时碰倒目标杯，而非受控持杯倾斜倒杯。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 56 的脚本以指尖逆解构形驱动肌肉及前臂，先执行接近抓杯动作，再按参考移动指尖；log step 60 确认该控制器实际执行了末回合。视频 0.12–0.40 s 显示手指动作未形成能带杯运动的接触，反而使杯子倾倒并与手分离，支持抓取执行上的控制失败；现有证据不足以确定更细的控制子类。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "无状态日志、接触对或接触力数据，也无成功示范；不能精确核验逐步位移和旋转误差，评分器未报告DAG所列四项原生指标，故其值均保留null。视频足以判断杯子被碰倒并滚离，但透明手部与低分辨率使具体承力手指和是否曾有瞬间对向夹持难以确定。agent 在 log step 62、64 自述杯子颜色掩码有偏差且修改发生在评分回合之后；缺少该回合物体真值和完整修改内容，不能据此独立确认感知根因或细分控制原因。未将20毫秒仿真控制周期直接换算为视频时间，所有视频锚点均取自所查帧。"
+ },
+ "summary_zh": "评分末回合仅执行14个控制步，完成倒杯轨迹节点未达成，最终为0/1。视频约 0.20 s，MyoHand手指动作碰倒杯子，杯子随后滚离手部；log step 60 记录末回合因杯子距参考或手腕过远而提前终止，根因仅能归到抓取执行的控制问题。无状态及接触力日志，无法进一步区分控制子类或核验精确追踪指标；评分后对脚本的修改不计入本次轨迹。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "末回合14个控制步后被服务判定失败，视频15帧与之吻合，杯子被碰倒滚离，节点未达成、断点与碰撞现象均有帧和服务读数支持；根因“控制”证据有限但已声明不确定。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:36:23Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--ayq8nmx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--ayq8nmx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0a84e728c5340f60f440057777a1e8b9013fa3533201878e1c20510b0eae00ef",
+   "video_sha256": "7c85ad9d9994f173f163c9ff5a3fcdaabcfb0bffbf68d8b730ca44325a54b626",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--odr8beb.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--odr8beb.json
new file mode 100644
index 0000000000000000000000000000000000000000..b7947f2b592036d3ccf5de378efc1fb2b6fa6650
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--odr8beb.json
@@ -0,0 +1,213 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-standard--odr8beb",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 22,
+   "sim_steps": 22,
+   "physics_steps": 220,
+   "steps": 49,
+   "calls": 7,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 49,
+   "elapsed_s": 3488.4,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.149,
+   "peak_processes": 12,
+   "peak_disk_gb": 0.371,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-standard--odr8beb/replay.mp4",
+  "duration_s": 0.92,
+  "fps": 25.0,
+  "frames": 23,
+  "sha256": "027afc17ef17b811e5d8f96aa6dbc0cb8f4a046af3b7a1f25ac8e30af0a3c2db"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.16,
+      "log_step": null,
+      "text_zh": "观察：已检查视频 0.00–0.88 s 的全部23帧及两侧视图；MyoHand手指从杯子侧上方抬起、张开并与杯子分离，杯子仍在桌面，未随手完成提起、倾斜和回落。视频 0.68–0.88 s 杯子仍独立立在桌面，手指在杯沿附近弯曲而未将其拿起。"
+     },
+     {
+      "t": 0.88,
+      "log_step": 62,
+      "text_zh": "观察：log step 62 的最终提交先重置，再向服务发送81行执行器动作；服务返回 episode=3、executed_rows=22、control_steps_chunk=22、done=True，以及物体距参考或手腕超过0.25米的失败原因。inputs/run.json 的评分字段为 graded_by_live=1、n_actions=22、success=0。推断：最后一回合未执行完要求的81个控制步，结合视频中未提杯的过程，完成倒杯轨迹节点始终未达到；此前被重置的两个回合不计入本次进展。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–0.16 s，MyoHand伸展的手指贴着杯子侧面及杯沿掠过，随后抬离；视频 0.80–0.88 s 手指弯向杯沿。未见拇指与其余手指形成夹住杯子的对向包覆，杯子未随手提起，因此实际记录为手指对杯子的拨动接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.16,
+   "log_step": 62,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.12–0.32 s，MyoHand向上离开杯子，杯子仍在桌面，没有被手带起；视频 0.88 s 仍未恢复抓持或完成倒杯。log step 62 的评分回合仅执行22个控制步，服务以物体距参考或手腕超过0.25米提前终止。推断：断点在倒杯轨迹操作中未建立可带杯运动的抓持，导航锚点取视频 0.16 s。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.16,
+    "log_step": 62,
+    "evidence_zh": "观察：视频 0.00–0.32 s 的连续帧显示MyoHand手指在杯侧伸展，随后空手上移；杯子始终留在桌面。视频 0.56–0.88 s 也未见手指夹住或托起杯子。推断：未建立的是带动杯子所需的有效抓持接触，虽有手指触碰杯侧，也不能算抓住后脱落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 0.16,
+    "log_step": 53,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：控制未能使手指闭合抓住杯子并与前臂抬升协调。log step 53 的提交脚本把预先生成的81×45执行器动作一次性按dt=0.02发送；log step 62 确认该脚本用于最终回合。视频 0.12–0.56 s 显示手已抬离、杯仍留在桌面，随后也未形成抓持；这支持控制层归因。现有证据不足以区分接触布局、肌肉激活与前臂跟踪误差各自的贡献，故不细分控制子类。"
+   }
+  ],
+  "gap_zh": "未提供状态日志或成功参考视频，手部局部存在遮挡，无法核验具体手指接触对与接触力。评分器reward字段及log step 62未报告track_pos_err、track_rot_err、track_ok或track_max，四项原生指标均留null；log step 60的本地优化分数不是最终评分器指标。22个20毫秒控制步对应仿真时长0.44 s，而视频时长为0.92 s；未建立精确时钟映射，所有定位直接使用所检查的视频帧，未将agent自述的仿真时间换作视频时间。",
+  "unmet": {
+   "cup_pour_tracked": false
+  }
+ },
+ "summary_zh": "评分采用log step 62重置后的第3回合，完成倒杯轨迹节点未达到，终态为0/1。视频 0.12–0.56 s 中MyoHand空手上移，杯子仍在桌面，所需抓持未建立；服务执行22个控制步后以物体距参考或手腕超过0.25米提前终止，证据支持控制层失败。缺少接触状态及最终追踪误差数值，无法细分控制根因或量出精确误差。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达到、final=0、断点 cup_pour_tracked、时钟（未把0.44 s仿真时间换算成视频时间）和 log step 62 的服务读数（22步、done、0.25 m 失败原因）与视频及 run.json 一致。没建立接触和控制根因有视频与服务读数支持，没有发现必须修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:10:44Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--odr8beb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--odr8beb/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "719cc53b1f26fc5abd8b12a0542c8612bc7a42c4da3a21c0ea710a6f577770f5",
+   "video_sha256": "027afc17ef17b811e5d8f96aa6dbc0cb8f4a046af3b7a1f25ac8e30af0a3c2db",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--plw9qmp.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--plw9qmp.json
new file mode 100644
index 0000000000000000000000000000000000000000..30bd2e23ec7dd3855c045dc427eb7946193f3ee4
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--plw9qmp.json
@@ -0,0 +1,213 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-standard--plw9qmp",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 20,
+   "sim_steps": 20,
+   "physics_steps": 200,
+   "steps": 58,
+   "calls": 16,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 58,
+   "elapsed_s": 1833.7,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.371,
+   "peak_processes": 19,
+   "peak_disk_gb": 1.126,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-standard--plw9qmp/replay.mp4",
+  "duration_s": 0.84,
+  "fps": 25.0,
+  "frames": 21,
+  "sha256": "1b47e5e06b2c4eea64652fea1600d7cee0163771d1c4dd4915a57e868607c811"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.28,
+      "log_step": null,
+      "text_zh": "观察：已检查视频全部21帧（视频0.00–0.80 s）。视频0.00–0.16 s粉色杯子在桌面，MyoHand手指伸在杯口和杯侧附近；视频约0.24–0.44 s手指弯曲，杯子偏移并倾倒，未见手指对向夹住杯子后将其提起。"
+     },
+     {
+      "t": 0.8,
+      "log_step": 165,
+      "text_zh": "观察：视频0.48–0.80 s杯子侧倒在桌面并与手逐渐分离，末帧仍未被拿起；没有完成提起、倾斜、回落的整段参考运动。log step 165记录agent脚本重置到episode 3后提交动作，服务在20个控制步返回done=True及“object more than 0.25 m from the reference (or from the wrist)”的失败原因；评分器字段n_actions=20、success=0，与未完成所需81控制步一致。该日志是最后评分回合，此前两回合不计入进度；没有把agent时间或控制步直接换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "推断：视频约0.20–0.44 s，MyoHand手指在杯口、杯侧处弯曲，杯子随之被拨偏并侧倒；按该连续运动判断为手指直接推拨杯子。未见对向夹持及持杯提起，故不标抓握；细小接触部位受透明手部模型和低分辨率限制。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.28,
+   "log_step": 165,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约0.24–0.44 s，MyoHand合指时杯子在桌面偏移、倾倒，没有被夹持提起；视频0.48–0.80 s杯子继续与手分离，直到末帧未修复。log step 165的最终episode 3在20个控制步因杯子距参考或手腕超过0.25 m而终止，未完成整段倒杯轨迹。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.28,
+    "log_step": 165,
+    "evidence_zh": "推断：视频0.00–0.80 s中，MyoHand手指从杯口附近弯曲后离开，杯子在桌面偏移并侧倒，始终没有随手被提起；所需的承载抓握从未建立。这里不否认短暂推拨接触，而是指没有建立能带杯追踪轨迹的夹持；没有已持住后再掉落的证据。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "log_step": 143,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 143写入的最终控制脚本把前臂目标直接设为IK计划前6列，同时仅将固定GRIP_VEC按k/10渐增来合指；log step 163保留这套抓取方式，log step 165在track模式运行最终评分回合。视频约0.24–0.44 s显示该合指方式拨偏、倾倒杯子而未形成对向承载夹持，因而手的运动不能传给杯子。会话中的本地诊断（log step 123、127）还记录预抓取位指尖位于杯周外围、初态杯子仅与桌面接触；这些仅支持接触布局推断，不作为评分回合的状态事实。没有经验证的可行替代握法，亦不能据这些失败试验断定任务不可行。"
+   }
+  ],
+  "gap_zh": "观察：inputs/中没有状态日志或成功示范；最终评分回合也未报告track_pos_err、track_rot_err、track_ok、track_max，四项原生指标保留null，不引用本地测试的误差冒充评分结果。视频显示杯子倾倒和手杯分离，但透明手部及分辨率不能给出精确接触对、接触力，也不能完全区分接触布局与肌肉执行偏差的贡献。视频文件时长0.84 s，最终服务记录20个20毫秒控制步；未建立两者的精确时钟映射，所有视频锚点均直接取自所查帧。"
+ },
+ "summary_zh": "观察：最终评分episode 3未完成唯一节点“完成倒杯轨迹”，节点最终为0/1；视频中MyoHand合指拨偏杯子，杯子侧倒在桌面并与手分离，log step 165记录20个控制步后原生失败终止。推断：固定合指方式未建立承载抓握，属于抓取和接触设计问题，未见持杯提起或完成倒杯参考运动。观察：没有评分回合状态日志、精确接触力或四项追踪指标，根因细分及精确误差仍有证据缺口。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达、断点、现象（没建立接触）和根因（抓取和接触设计，alt_feasible 未知）与视频帧及 log step 165 一致，视频时间未换算自 agent 时间，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:10:06Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--plw9qmp/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--plw9qmp/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "eccc0348397f699544a9903299d64b354b7eee6d1e5372a69674e0532ceaf34f",
+   "video_sha256": "1b47e5e06b2c4eea64652fea1600d7cee0163771d1c4dd4915a57e868607c811",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--pvfxfhw.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--pvfxfhw.json
new file mode 100644
index 0000000000000000000000000000000000000000..2e3a110092f4a13268539fa9aaed214964a32e04
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--pvfxfhw.json
@@ -0,0 +1,213 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-standard--pvfxfhw",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 16,
+   "sim_steps": 16,
+   "physics_steps": 160,
+   "steps": 111,
+   "calls": 85,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 111,
+   "elapsed_s": 868.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.447,
+   "peak_processes": 7,
+   "peak_disk_gb": 0.254,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-standard--pvfxfhw/replay.mp4",
+  "duration_s": 0.68,
+  "fps": 25.0,
+  "frames": 17,
+  "sha256": "9e3801712a3530c3f1ecc9fb14bb6ef0141dcb6143704eada6186718af1231c1"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.16,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.64 s 的全部17帧；视频 0.12–0.28 s，MyoHand手指弯曲靠拢，粉色杯子在手边倾倒并向画面下方滑移；视频 0.32–0.64 s，手指空合，杯子与手分离并移出视野，未出现抓稳后随手提起、倾斜及回落的完整倒杯轨迹。"
+     },
+     {
+      "t": 0.64,
+      "log_step": 131,
+      "text_zh": "观察：log step 131 显示agent脚本重置至第11回合后提交81个控制步，服务仅执行16步便报告 done=True，原因是物体距参考位置或手腕超过0.25米；log step 132 确认第11回合已结束，评分器 reward.sim_steps=16、success=0。推断：该末回合未满足完整执行81步的必要条件；此前重置掉的调试回合不计入本节点进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.12–0.28 s，MyoHand手指合拢时在粉色杯子一侧推碰杯壁，杯子倾倒并离开手边；未见拇指和其余手指形成可带动杯子的对向夹持，实际接触按推碰记录。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.16,
+   "log_step": 131,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.16 s，机器人合指时杯子开始明显偏离手边并倾倒，视频 0.28–0.64 s 杯子与手分离，未重新抓回。log step 131、132 的服务结果确认评分末回合只执行16个控制步即因物体距参考位置或手腕过远而失败；断点属于正在尝试的完整倒杯轨迹步骤。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.16,
+    "log_step": null,
+    "evidence_zh": "推断：视频 0.00–0.12 s 杯子仍在原支撑处，视频 0.16–0.28 s 手指弯曲时杯子倾倒并滑离，视频 0.32–0.64 s 手指空合；整个过程中未见杯子被对向夹住后随手运动。因此未建立本步骤所需的承载抓握，虽有推碰杯壁，不能据此认定曾经抓稳后脱手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 0.16,
+    "log_step": 130,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 130 的最终脚本以预设的拇指、屈指及腕部肌肉激励逐步合指，并一次提交整段动作；log step 131 确认该脚本执行的是评分末回合。结合视频 0.12–0.32 s 合指把杯子推离、未形成带杯运动的抓握，可将失败归于抓取接触的控制未奏效；现有证据不足以进一步区分接触几何、肌肉激励参数或动态跟踪问题。"
+   }
+  ],
+  "gap_zh": "未提供状态日志、接触对或接触力，透明手部与杯壁有局部遮挡，无法核实各手指的精确接触时刻及力；未见承载夹持，故不推定曾抓稳。评分器未报告 track_pos_err、track_rot_err、track_ok、track_max，四项数值均保留null；agent在此前调试中的运动学预测不作为评分回合的实测指标。无成功参考视频。视频时刻直接取自25 fps的帧位置，未把50 Hz控制步或agent工作时间换算为视频时间。"
+ },
+ "summary_zh": "评分末回合中，机器人尝试合指抓杯，但杯子被推碰后倾倒、滑离，未建立能带杯执行倒杯运动的抓握，完成倒杯轨迹节点始终未达成（0/1）。log step 131、132 确认第11回合只执行16个控制步便因物体距参考位置或手腕过远而终止；归因停留在控制，缺少接触力、状态日志及四项原生追踪指标，无法细分。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点、现象与根因均与视频（17帧：杯子被合拢的手指推倒并滑离）及 log step 131/132（评分回合16步即因物体距参考超0.25 m终止）一致；时钟使用视频帧位置，未换算agent时间。未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:38:33Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--pvfxfhw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--pvfxfhw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "4c06efb0cb9077e78cb856dec00697a8cd79a68d67508bfe431c173c102cb1ee",
+   "video_sha256": "9e3801712a3530c3f1ecc9fb14bb6ef0141dcb6143704eada6186718af1231c1",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--rnojs3x.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--rnojs3x.json
new file mode 100644
index 0000000000000000000000000000000000000000..b22fc6975ebb2e45347b0b23be82cfb649115050
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--rnojs3x.json
@@ -0,0 +1,216 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-standard--rnojs3x",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 58,
+   "sim_steps": 58,
+   "physics_steps": 580,
+   "steps": 594,
+   "calls": 209,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 594,
+   "elapsed_s": 3407.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.485,
+   "peak_processes": 18,
+   "peak_disk_gb": 1.208,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-standard--rnojs3x/replay.mp4",
+  "duration_s": 2.36,
+  "fps": 25.0,
+  "frames": 59,
+  "sha256": "b5e607b7cfbf9273a3222c354a4ba0dcb6099c52100da3c85520bc5753106f7a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.28,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.64 s，MyoHand手指围合粉色杯子；视频 0.80–1.60 s，杯子随手离开桌面并倾斜，表明确有抓持和倒杯运动尝试，但这不是整段参考轨迹完成。"
+     },
+     {
+      "t": 2.12,
+      "log_step": null,
+      "text_zh": "观察：视频 1.76 s 杯子仍在手指间，视频 2.04–2.16 s 杯子从手指下方滑脱并下落；视频 2.32 s 手中已空，未重新抓回或完成回落。"
+     },
+     {
+      "t": 2.32,
+      "log_step": 135,
+      "text_zh": "观察：log step 135 的服务结果明确是最后第11回合，control_steps=58、done=True，原因为物体距参考位置或手腕超过0.25米；inputs/run.json的reward也报告n_actions=58、live_success=0、graded_by_live=1。该回合未执行完节点要求的81个控制步，与视频末尾脱手一致。控制步只用于确认回合和提前终止，不换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.64–1.76 s，单只MyoHand的拇指与其余手指从杯体两侧围合抓持粉色杯子，带动其提起、倾斜；视频约 2.12 s 抓持失去，杯子下落。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 2.12,
+   "log_step": 135,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 2.12 s，正在进行倒杯运动的MyoHand未能继续夹持杯子，杯子向下离开手指；视频 2.32 s 手中为空，直到结束未修复。log step 135确认评分所用第11回合在第58个控制步因物体距参考位置或手腕超过0.25米提前失败，整段倒杯轨迹未完成。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.12,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.80–1.76 s，粉色杯子被MyoHand围合抓持并随手抬起、倾斜；视频 2.04–2.16 s，杯子逐渐从手指下方脱离并下落，视频 2.32 s 手中为空。此前已建立承载抓持，后续丢失，没有重新抓回。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 133中agent的最终执行脚本把两套执行器序列按0.75和0.25混合，随后通过一次submit整批执行，期间没有根据新观测修正命令；log step 135的FINAL BASIS 6 5和第11回合结果确认执行了这一最终序列。视频 1.76–2.32 s显示手指与杯子的承载关系在运动中失去，说明该执行器控制序列未能维持抓持。现有证据不足以细分为接触几何、轨迹求解、反馈或某个肌肉参数问题。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "没有状态日志、接触对或接触力记录，也没有成功示范；最后评分回合未报告track_pos_err、track_rot_err、track_ok和track_max，四项原生指标均保留null，不能用第5或第6测试回合的数值替代。视频可确定抓持后脱手及未完成，但无法确定更早首次超出精确误差阈值的时刻或抓持失效的控制子类；所有导航时刻均直接取视频帧，未将控制步或agent工作时间换算为视频时间。"
+ },
+ "summary_zh": "评分采用最后第11回合：MyoHand抓住杯子并提起、倾斜，但未完成整段倒杯参考轨迹，节点最终为0/1。视频约 2.12 s 杯子脱手下落，log step 135确认第58个控制步因距离超限提前失败；根因仅能归到控制序列未维持抓持。没有状态日志及最后回合的精确追踪指标，无法进一步定位首次误差超限或控制子类。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "标签与视频、log一致：评分回合（第11回合，log step 135）杯子被抓起、倾斜后约2.0–2.3 s脱手下落，节点未完成；断点、接触、现象与根因（标为推断且承认证据不足）均有依据，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:36:35Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--rnojs3x/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--rnojs3x/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "12dd1fe94719f210185b822e1db4f8235ca973a18d6666941c5b3d9fbfdfce22",
+   "video_sha256": "b5e607b7cfbf9273a3222c354a4ba0dcb6099c52100da3c85520bc5753106f7a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--ugtexpt.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--ugtexpt.json
new file mode 100644
index 0000000000000000000000000000000000000000..29fb0df866625e81030f163c71ea53ab41d15e65
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--ugtexpt.json
@@ -0,0 +1,211 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-standard--ugtexpt",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 72,
+   "calls": 15,
+   "episodes": 5,
+   "resets": 4,
+   "reset_budget": 10,
+   "sim_steps_total": 72,
+   "elapsed_s": 2314.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.819,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.438,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-standard--ugtexpt/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "c5d8ec1b609188a1b6138b8f8f4ec6a4d9397e8d488e4e7afacf381954dc4de0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.28,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.28 s，MyoHand单手的拇指与其余手指环绕粉色杯体；随后视频 0.36–0.60 s，杯子逐渐倾倒、与手指分离，绿色球从杯内移出，手没有带着杯子完成提起、倒杯及回落。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 231,
+      "text_zh": "观察：视频 0.64–0.76 s，杯子向画面下方移出，灵巧手空置，未重新抓回；log step 231 的机器人服务返回第5回合仅执行19个控制步便done=True，原因为物体距参考位置或手腕超过0.25米，与reward的n_actions=19、episodes=5、graded_by_live=1一致。此为最后实时回合，后续log step 234–236只做本地设计和文件交付，未再执行机器人动作；完整81步轨迹未完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.00–0.28 s，MyoHand单手的拇指与其余手指从杯体两侧环绕夹持粉色杯子；视频 0.36–0.60 s 杯体从指间滑离，此夹持未能持续承载杯子。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.4,
+   "log_step": 231,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.36–0.60 s，机器人手指与杯体的包围关系逐渐丢失，杯子倾倒在桌面上而手继续移开；至视频 0.76 s 没有重新抓回。log step 231 的最后实时回合由服务提前判失败，仅执行19个控制步，未完成81步倒杯轨迹。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 0.4,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.28 s，MyoHand手指从两侧环绕杯子；视频 0.36–0.60 s，杯体从指间向外滑转、倒在桌面，手指随后空置。视频 0.64–0.76 s 杯子与手持续分离，未恢复夹持。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 230–231 显示agent在本地模型中生成肌肉与前臂的整段命令，再一次性以hold方式提交给机器人，执行中未依据实际杯体滑动修正命令；视频 0.36–0.76 s 显示实际夹持丢失后手继续离开杯子，说明该肌肉与前臂控制序列未能维持带杯运动。log step 234及237中agent也自述夹持在过渡阶段失效；本地32/81结果是其自建模型的测试，不是本回合评分。证据不足以进一步区分接触布局、肌肉激活或轨迹跟踪问题，保留在控制大类。"
+   }
+  ],
+  "gap_zh": "未提供状态日志、接触对或接触力，不能定量区分抓握几何、肌肉激活与前臂跟踪各自的贡献；评分器未报告track_pos_err、track_rot_err、track_ok、track_max，四项数值均留null。视频时钟直接按所检帧标注；19个20毫秒控制步与0.8秒视频时长不相等，未把控制步或agent工作时间换算为视频时间。无成功参考视频。"
+ },
+ "summary_zh": "最后实时回合未完成唯一节点“完成倒杯轨迹”，最终进度0/1；机器人手指起初环绕杯体，视频约0.40 s起杯子滑转脱离，随后倒在桌面而手空置移开。log step 231的服务返回19个控制步后原生失败终止，结合视频可将失败归于未维持夹持的控制问题。缺少状态、接触力及实际追踪误差字段，无法细分控制根因或给出四项原生指标数值。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频一致：手起初环抱杯子，约0.4 s起杯子滑脱倒在桌面，节点未达成，breakpoint 与「没保持住」、根因控制（未细分）均有证据支持，时钟未换算。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:37:47Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--ugtexpt/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--ugtexpt/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9b59f1925cac91aa7b7e9a5a914256797dcc26050e0d73764eb43cbd29371920",
+   "video_sha256": "c5d8ec1b609188a1b6138b8f8f4ec6a4d9397e8d488e4e7afacf381954dc4de0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--zplxqah.json b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--zplxqah.json
new file mode 100644
index 0000000000000000000000000000000000000000..28332c9845aa761cbc74ac339c616588594d7af8
--- /dev/null
+++ b/phase2/myosuite/dex_cup_pour/trajectories/myosuite-dex-cup-pour-i00-standard--zplxqah.json
@@ -0,0 +1,240 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cup-pour-i00-standard--zplxqah",
+ "taskKey": "myosuite/dex_cup_pour",
+ "dag_sha256": "8e77d2cf6ffaa66127634b52432fa68a93f26c48d76a25065fe8795dc5cfd63c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 22,
+   "sim_steps": 22,
+   "physics_steps": 220,
+   "steps": 183,
+   "calls": 116,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 183,
+   "elapsed_s": 2297.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.437,
+   "peak_processes": 20,
+   "peak_disk_gb": 0.505,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cup-pour-i00-standard--zplxqah/replay.mp4",
+  "duration_s": 0.92,
+  "fps": 25.0,
+  "frames": 23,
+  "sha256": "9e44f5ac96296153b7d4795e75f7bb51bc9427680afa2463838cd9968097053e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cup_pour_tracked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 双视图中粉色杯子位于桌面，MyoHand手指位于杯旁；初始状态不是完整倒杯轨迹的完成状态。"
+     },
+     {
+      "t": 0.28,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.88 s 的全部23帧，MyoHand手指在杯旁收拢，杯子向手外偏移，随后倾斜翻倒；视频 0.40 s 右视图仍见绿球在杯内，视频 0.64 s 右视图已见杯底及离杯的绿球，视频 0.88 s 杯子翻倒、绿球已离开画面。整段未见杯子随手离桌提起并完成参考倒杯运动。"
+     },
+     {
+      "t": null,
+      "log_step": 64,
+      "text_zh": "观察：log step 64 的服务返回先列出旧回合episode=10，再显示reset 11 0；最后回合episode=11在control_steps=22时done=True，原因为杯子距参考或手腕超过0.25米。run.json评分器字段n_actions=22、episodes=11、resets=10、live_success=0与此一致，未完成DAG要求的81控制步。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "推断：视频 0.04–0.28 s，MyoHand手指在杯壁旁收拢时杯子向外偏移，随后杯子被推倾翻倒；视频 0.64–0.88 s 绿球离杯、杯子与手指逐渐分离，实际交互为手指推拨杯子，未见能使杯子随手提起的稳定抓握。低分辨率画面不足以确定具体接触指节。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 73,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cup_pour_tracked",
+   "t": 0.28,
+   "log_step": 64,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.28 s，MyoHand手指已在杯旁收拢，杯子却向手外偏移；视频 0.64–0.88 s 杯子倾斜翻倒、绿球离杯，未见杯子随手提起，整段参考倒杯运动未完成。log step 64确认评分的最后回合episode=11只执行22控制步便因杯子距参考或手腕超过0.25米而结束。"
+  },
+  "unmet": {
+   "cup_pour_tracked": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.28,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.28 s 杯子在MyoHand手指收拢时向手外偏移；视频 0.40–0.88 s 右视图中杯子由杯口可见绿球转为倾斜翻倒、露出杯底，绿球掉出杯外。杯子未随手提起并完成参考倒杯运动。推断：实际杯子位姿运动偏离所需轨迹，而非完全没有物理接触；未见先被稳定持住再滑脱的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 64,
+    "evidence_zh": "推断：视频 0.04–0.88 s 中MyoHand手指收拢后杯子被推倾翻倒，未形成能使杯子随手提起的稳定抓握；log step 64的服务返回确认最后回合episode=11在control_steps=22时因杯子距参考或手腕超过0.25米而终止。据此将根因保留在宽泛的“控制”：所执行的手部与前臂命令未实现提杯和轨迹带动。log step 42及58仅说明最终执行预生成命令的开环回放，不证明回放脚本、参数或反馈机制本身有错；log step 65的不可达解释是agent自述，不作为根因证据。可行替代方案为未知：本目录无成功示范，也无独立验证其他合法抓握及控制方案的证据，不能据此次失败排除全部方案。",
+    "source_zh": "日志 / 视频",
+    "alt_feasible": "未知"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功示范；视频分辨率较低，可判杯子未被提起，但不能精确确定接触指节及接触力。评分器reward未报告track_pos_err、track_rot_err、track_ok、track_max，四项原生指标均填null；log step 64的离线仿真误差不是评分回合的原生指标。最后回合22个20毫秒控制步与视频0.92 s的时长不直接相等，故未把控制步换算为视频时刻，全部视频锚点取自已检查帧。agent在log step 65提出负载和旋转不可达的解释，未有排除全部合法方案的独立证据，不据此归为任务或环境问题。"
+ },
+ "summary_zh": "评分采用最后一次重置后的第11回合，完成倒杯轨迹节点最终为0/1：视频 0.00–0.88 s 中MyoHand手指收拢，杯子向外偏移后倾斜翻倒，绿球离杯，未随手提起。失败位于完成倒杯轨迹的操作中，现象改为位姿不准；控制根因仅据实际回放未能稳定抓握和带动杯子的结果推断，log step 64确认最后回合在控制步22提前终止。无状态日志、成功示范和未舍入追踪指标，不能细分控制根因或判定是否存在可行替代方案。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "视频 0.00–0.88 s 右视图显示杯子被推得倾斜、杯口转向镜头并翻倒，里面的绿球掉出，而不是单纯沿桌面平移；标签的观察和接触现象与此不符。节点未达成、断点 cup_pour_tracked 操作中均正确。",
+    "n_issues": 3,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点（cup_pour_tracked，操作中）、现象位姿不准、根因控制均与视频一致，上一轮意见已采纳，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:10:21Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--zplxqah/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--zplxqah/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--zplxqah/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cup-pour-i00-standard--zplxqah/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b278e5b80cea9e00d9d65e9f24e9b728823bfd214e34273bb0b4f1ca1c2d748f",
+   "video_sha256": "9e44f5ac96296153b7d4795e75f7bb51bc9427680afa2463838cd9968097053e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--3wrqpfe.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--3wrqpfe.json
new file mode 100644
index 0000000000000000000000000000000000000000..c6335292b637024507198b42cfe29f6d6c3acd8c
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--3wrqpfe.json
@@ -0,0 +1,195 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-privileged--3wrqpfe",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 86,
+   "sim_steps": 19,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 63,
+   "agent_resets": 42,
+   "agent_sim_steps": 277,
+   "agent_records": 0,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.754,
+   "peak_processes": 11,
+   "peak_disk_gb": 1.002,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-privileged--3wrqpfe/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "fca07f4ab48d696290384e9391602827c94ee463db4f32c6fa85a8b07dcdb5bb"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的俯视和侧视中，红色小圆柱在桌面上，单只 MyoHand 的手指位于其一侧；此时尚未执行完整参考运动。"
+     },
+     {
+      "t": 0.32,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.76 s（全部20帧），视频 0.24–0.56 s 手指屈曲并从圆柱一侧碰拨，圆柱仅在桌面上小幅移位；视频 0.60–0.76 s 手指离开，圆柱仍在桌面，没有随手抬起、前送的过程。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 136,
+      "text_zh": "观察：log step 136 对最终保存动作的 Session.replay 返回 success=False、steps=19、done=True、track_pos_err=0.28446 m、track_ok=13，原生终止原因为圆柱距参考或手腕超过0.25 m；log step 138 对同一交付文件的两次新进程回放均在19个控制步结束且末态哈希一致。视频末帧 0.76 s 圆柱仍在桌面，与未完成全部86个控制步的回放结果一致；这里不将控制步乘0.02秒作为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.24–0.56 s，单只 MyoHand 的屈曲手指在红色圆柱一侧碰拨，使其在桌面上小幅移位；未见拇指与其余手指对置夹持圆柱，视频 0.60–0.76 s 手离开而圆柱留在桌面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": 0.28446
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.6,
+   "log_step": 136,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.24–0.56 s，MyoHand 在圆柱旁屈指并碰拨；到视频 0.60 s，手指开始与圆柱分离，直至视频 0.76 s 圆柱仍留在桌面，未被携起。log step 136 的最终交付回放在19个控制步触发原生距离终止，log step 138 的两次新进程回放复现同一失败，整段圆柱循迹交接未完成。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.6,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.24–0.56 s 可见手指从红色圆柱一侧碰拨，但没有形成包围或对置夹持；视频 0.60–0.76 s 手指空手离开，圆柱仍在桌面。推断：所需的携物抓握从未建立，已有的单侧碰触没有将圆柱拿起，不是抓住后脱手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent 的肌肉驱动控制没有实现抓握所需的手指运动。log step 134、136 记录最终控制器将参考关节位姿经惯性缩放PD和非负最小二乘转换为肌肉指令，并叠加屈指激励；此前会话测试 log step 116 实测手关节跟踪RMS为0.3169 rad，log step 132 的静态抓握姿态测试也残留0.313–0.388 rad误差。这些是会话诊断而非评分轨迹状态，但与视频 0.24–0.76 s 手指屈曲、单侧碰拨却未抓起圆柱的评分行为一致；证据支持控制层面的失败，尚不足以区分接触设计、轨迹或反馈控制子类。"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功参考视频；两视角足以确认没有拿起圆柱，但不能确定每根手指的精确接触对及受力。log step 136 提供了最终交付在原生终止时的 track_pos_err=0.28446 m（并非第86步末步误差），仅报告 track_ok=13，未直接报告 track_ok_fraction 或 track_max，故后两项原生指标为null。会话中自行延长到86步的诊断不是评分回放，未用于节点进度或最大误差。agent 在 log step 139 对参考动力学不可行的解释不足以排除所有合法方案，无法据此确定任务或环境根因。"
+ },
+ "summary_zh": "评分轨迹完成0/1个节点：MyoHand 屈指并单侧拨动红色圆柱，但未建立可携物的抓握，视频 0.60–0.76 s 手离开时圆柱仍在桌面。最终交付的86条动作在两次新进程回放中均只执行19个控制步即触发原生距离终止；会话跟踪诊断支持控制层面的失败。无状态日志和成功示范，无法精确确定接触力及控制子类，完整轨迹覆盖率与最大误差也未报告。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、接触（推/扫）、断点（操作中）与 控制 根因（停在 item 控制）均与视频一致；时间均为视频时钟，未把控制步换算成视频时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:40:17Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--3wrqpfe/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--3wrqpfe/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "36023119ad58fab718a2fd6486d1e0b28fe686cf6fca1e5790b63618db82ee1c",
+   "video_sha256": "fca07f4ab48d696290384e9391602827c94ee463db4f32c6fa85a8b07dcdb5bb",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--fkncezy.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--fkncezy.json
new file mode 100644
index 0000000000000000000000000000000000000000..57149de7c9ed42d737c7ce5eff45b59ab5be68b7
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--fkncezy.json
@@ -0,0 +1,195 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-privileged--fkncezy",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 52,
+   "agent_resets": 79,
+   "agent_sim_steps": 213,
+   "agent_records": 2,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.71,
+   "peak_processes": 17,
+   "peak_disk_gb": 0.277,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-privileged--fkncezy/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "d4e33bdce755f083b3fb75255dd87cdba8a37ec95af408a085f1cd18216b7dfd"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.76 s 的全部20帧及两视角显示，MyoHand在红色圆柱旁屈指、随后抬离，红色圆柱始终留在桌面；视频 0.56–0.76 s 圆柱与升起的绿色标记明显分离，未出现随手升起、前送的过程。"
+     },
+     {
+      "t": 0.56,
+      "log_step": 54,
+      "text_zh": "观察：log step 52 保存交付轨迹，log step 53 的两次新进程回放均报告19行、19控制步、success=false、相同末态哈希；log step 54 重新生成同一交付文件，报告原生done=True、track_pos_err=0.28932 m、track_ok=13，并再次回放失败。评分器reward的n_actions=19、sim_steps=19、deterministic=1与这些回放相符。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 53,
+      "text_zh": "推断：视频未见圆柱离桌循迹，且交付回放在19个控制步即因圆柱距参考或手腕过远而终止，不能完成DAG要求的全部86步；本事件节点全程未达成。视频时间直接取自逐帧检查，未将控制步或agent时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "推断：视频 0.36–0.52 s 两视角中，单只MyoHand屈曲的指端贴到红色圆柱侧缘，表现为短暂单侧轻触/压碰；未见拇指与其余手指在圆柱两侧形成夹持，圆柱也未随手离桌。透明手部及像素分辨率限制使具体接触指和压力不能确定。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.56,
+   "log_step": 53,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.56 s，MyoHand已屈指并向上离开圆柱，红色圆柱仍在桌面，绿色标记开始明显升离；到视频 0.76 s 仍未带起圆柱。log step 53 对实际交付文件的两次回放都在19个控制步报告“object more than 0.25 m from the reference (or from the wrist)”并失败；该循迹步骤此后未修复。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.56,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.76 s，两视角均未见MyoHand从两侧夹住红色圆柱；视频 0.36–0.52 s 指端虽靠贴圆柱侧缘，但视频 0.56–0.76 s 手升离时圆柱仍由桌面承托，未随手运动。推断：未建立能带动圆柱的抓握接触，并非先抓持再滑落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 52 的交付生成脚本以手内固定点跟踪参考位移，同时用肌肉控制手指闭向参考握姿，未根据圆柱是否已被夹持切换动作；log step 54 的同一脚本执行及回放仍在19控制步终止。结合视频 0.36–0.76 s 手指在圆柱侧边屈曲、手离开而圆柱留桌，执行控制未建立带动物体所需的抓持；现有证据不足以细分为接触几何、轨迹求解或反馈参数问题。"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功参考视频；透明手部和低分辨率使具体接触指、压力及是否存在极短暂的对向接触无法精确核实，但圆柱未被带离桌面清楚可见。reward未报告DAG三个原生指标的数值，故native_metrics均留null；log step 54虽报告交付生成运行末步track_pos_err=0.28932 m、track_ok=13，未报告track_max或覆盖率，未据此重建评分器指标。agent在log step 57声称隐藏质量和动力学使实例不可解，现有会话测试不足以排除所有合法方案，不能据此认定任务设计问题；控制根因仅判到大类。"
+ },
+ "summary_zh": "交付轨迹中MyoHand在圆柱旁屈指并抬离，圆柱始终留在桌面，完整循迹交接节点未达成（0/1）。视频约 0.56 s 开始明显分离，交付回放在19个控制步触发距离终止；未建立带动物体的抓握，根因保留为控制。缺少状态日志，具体接触与控制子因不能精确确定，也不能证实agent所称实例不可解。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "复核：视频20帧两视角中圆柱始终留在桌面，绿色参考标记升离，节点未达成；log step 53/54回放19步即终止，与标签一致。断点、没建立接触、控制根因均有证据支持，未发现需修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:36:05Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--fkncezy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--fkncezy/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "561d01f1066f91545b848ffeb51b851ba25a5d76e3ca5f5c5bf63c19425658a8",
+   "video_sha256": "d4e33bdce755f083b3fb75255dd87cdba8a37ec95af408a085f1cd18216b7dfd",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--gwjdrkv.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--gwjdrkv.json
new file mode 100644
index 0000000000000000000000000000000000000000..547f705a83f438b106a37f16ea9f9e142b7b9ef4
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--gwjdrkv.json
@@ -0,0 +1,193 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-privileged--gwjdrkv",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 51,
+   "agent_resets": 256,
+   "agent_sim_steps": 2153,
+   "agent_records": 0,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.824,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.242,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-privileged--gwjdrkv/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "6e1a81c17d8173781c5eb34221daeba61a65756acd72e3c6445e92be7ca9582c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.76 s 的全部20帧；两视角中，MyoHand手指在圆柱旁上方弯曲，随后手抬离桌面，红色圆柱始终留在桌面，未随手升起或前送。视频 0.24 s 可见合指时手与圆柱之间仍有间隙。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 56,
+      "text_zh": "观察：视频 0.76 s 末帧中，圆柱仍在桌面，手已抬离；log step 56 对已保存trajectory.npz的两次独立回放均报告steps=19、rows=19、success=false，终止原因为物体距参考或手腕超过0.25米。评分器reward同样给出sim_steps=19、task_failure=1、deterministic=1，因此完整86步循迹从未完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频 0.00–0.76 s，单只MyoHand手指在圆柱旁上方合拢并抬离，未包住或夹住圆柱。候选方式“未触物的合指尝试”：现有抓握、推/扫、压/擦等均要求实际作用于物体，不适用于此次未建立目标接触的尝试；按接触规则保留尝试，不能记为已发生抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.24,
+   "log_step": 56,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.24 s，机器人已在圆柱旁合指，但手指没有夹住圆柱；此后至视频 0.76 s 手继续抬离，圆柱始终留在桌面，抓取未获修复。log step 56 的提交文件回放在控制步19提前失败，未完成所需86步。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.24,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.76 s 的两视角均未显示手指包夹圆柱；视频 0.24 s 手在圆柱上方合拢，视频 0.40–0.76 s 手与桌上圆柱的间距继续扩大，圆柱没有被带离桌面。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 56,
+    "evidence_zh": "推断：log step 55 的手臂偏置及闭指时刻试验均在控制步19失败；log step 56 写出的控制程序以参考手臂位置、超前项和偏置驱动手臂，并按预定时刻增加屈指激励，提交轨迹的两次回放仍均失败。结合视频 0.24–0.76 s 合指发生在圆柱旁上方且未带动物体，控制未能把手指闭合与到达圆柱位置协调起来；证据不足以进一步区分接触构型、轨迹跟踪或具体参数问题。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功示范，无法精确重建接触距离、跟踪误差及力学失配原因；正式reward未报告track_pos_err、track_ok_fraction、track_max，三项原生指标留null。log step 59及之后的距离、接触和力学诊断属于会话测试，不能作为提交回放的逐步状态；log step 66自述交付为86行，但log step 56实测提交文件rows=19，两次回放及正式reward均为19步，采用后者。不将仿真控制步时长或agent工作时间换算成视频时间。"
+ },
+ "summary_zh": "灵巧手在圆柱旁合指并抬离，但圆柱始终留在桌面，“圆柱循迹交接”未达成，最终0/1；提交回放在控制步19提前终止。断点为操作中的未建立接触，归因到控制未能协调到位与闭指；缺少状态日志及正式跟踪指标，无法确认更细的力学原因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与证据一致：视频0.00–0.76 s（20帧）手在圆柱旁抬离，圆柱始终留在桌面；节点未达成、断点与“没建立接触”、根因“控制”（推断，未细分）均有日志和评分器数字支撑，无时钟换算问题。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:39:49Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--gwjdrkv/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--gwjdrkv/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "86ca444371b920524d78e5c25eced93a20d854e6dd6a2872e5ce27c39db62cb4",
+   "video_sha256": "6e1a81c17d8173781c5eb34221daeba61a65756acd72e3c6445e92be7ca9582c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug.json
new file mode 100644
index 0000000000000000000000000000000000000000..8eb5cc3f9dbb13458ddefd1886deccdfcfa63b3e
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug.json
@@ -0,0 +1,207 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-privileged--rrd9vug",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 86,
+   "sim_steps": 15,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 248,
+   "agent_resets": 1117,
+   "agent_sim_steps": 3462,
+   "agent_records": 121,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug/replay.mp4",
+  "duration_s": 0.64,
+  "fps": 25.0,
+  "frames": 16,
+  "sha256": "4a5e8b85901f396bf3193450c110b0d1e09f24a500e8dffe6b3b538e1e5ed40e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.60 s 的全部16帧。红色小圆柱起初直立在桌面；灵巧手向其附近运动，但未见手指从相对两侧合拢夹持圆柱，也未见圆柱随手离开桌面。"
+     },
+     {
+      "t": 0.56,
+      "log_step": null,
+      "text_zh": "观察：视频 0.48–0.60 s，红色圆柱仍在桌面附近，末段倾斜并与绿色参考标记分离，未出现要求的升起、前送和回落过程。"
+     },
+     {
+      "t": null,
+      "log_step": 63,
+      "text_zh": "观察：log step 63 的控制脚本输出明确保存交付轨迹，报告执行15步即失败；run.json 的评分器字段也为 sim_steps=15、n_actions=86、task_failure=1、replay_success=0。推断：结合视频中未抬起的圆柱，完整86步循迹节点未达成；会话后续的抓取扫描和控制器测试不计入该节点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–0.60 s，单只MyoHand灵巧手向小圆柱附近运动，圆柱末段倾斜，未见包夹抓握。推断：画面尺寸及透明叠影不足以确定是否发生短暂指尖碰触及其具体方式，因此不将典型抓握方案记作实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.48,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.40–0.60 s，灵巧手未夹起圆柱，圆柱留在桌面并在末段倾斜，整段循迹未启动有效的持物运动；评分器记录实际只执行15个仿真控制步便 task_failure=1。推断：断点为正在尝试的圆柱循迹交接，视频 0.48 s 为未建立抓握、圆柱仍留桌面的导航锚点。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.48,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.60 s，灵巧手虽向圆柱附近运动，但圆柱未被相对手指包夹、未随手抬离桌面，末段反而倾斜。推断：没有建立搬运所需的有效抓握接触；可能的短暂碰触不等于曾握住后丢失。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": null,
+    "log_step": 63,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "观察：log step 56 的 gen6.py 以参考关节轨迹、逆动力学及肌肉激活求解驱动前臂和手指；log step 63 对其加入腕部位移补偿后保存了仍在15步终止的候选。视频 0.00–0.60 s 显示该尝试未形成包夹、圆柱未被抬起。推断：脚本的目标包含接近、闭指和循迹，但执行未实现有效抓取，归因于控制；现有证据不能进一步区分接触构型、跟踪偏差或肌肉控制细节。"
+   }
+  ],
+  "gap_zh": "观察：inputs/ 中没有状态日志或成功参考视频；评分器 reward 未报告 track_pos_err、track_ok_fraction、track_max，因此 native_metrics 均留空值。视频仅16帧，接触区域小且存在透明叠影，不能可靠确定短暂接触的手指与方式，也不能量化逐步误差或确认原生终止的具体距离分量。log step 63 的保存候选与评分器均执行15步，但没有交付文件哈希可进一步核对；后续会话测试的数值不作为评分回放指标。"
+ },
+ "summary_zh": "评分回放未完成唯一的圆柱循迹交接节点，最终为0/1：视频 0.00–0.60 s 中圆柱始终未被抓起，末段倾斜，评分器记录15步即失败。断点在操作中，现象为未建立有效抓握，根因保留为控制；缺少状态日志，无法确定短暂指尖接触及更细的控制失效机制，原生跟踪指标也未由评分器报告。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点、现象与视频一致（16帧：圆柱留在桌面，绿色参考标记升起后分离）；但 progress 证据和根因证据声称「log step 63 保存了交付轨迹并执行15步」，log 并不支持：step 63 只是网格搜索，输出无 saved，交付文件是哪一版无法从日志确认。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、没建立接触现象与视频一致（16帧中圆柱留在桌面，末段倾斜，绿色参考标记升起后分离）；round 1 的质疑已被日志反证：log step 63 输出含 SAVED /app/output/trajectory.npz，且与评分器 sim_steps=15 相符。控制根因表述保守，可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:10:06Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-privileged--rrd9vug/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f0ac287566eb24c3c17d1c0baceb9cf8ff561282f913bc657d55ff0339a6a971",
+   "video_sha256": "4a5e8b85901f396bf3193450c110b0d1e09f24a500e8dffe6b3b538e1e5ed40e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--fynmua5.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--fynmua5.json
new file mode 100644
index 0000000000000000000000000000000000000000..315898de9c562475e95600d8700d90adefacd792
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--fynmua5.json
@@ -0,0 +1,211 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-standard--fynmua5",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 169,
+   "calls": 136,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 169,
+   "elapsed_s": 3004.5,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.325,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.044,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-standard--fynmua5/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "26042e16438c2a3c8c235563f9af817290f655424eef4d757dd28bf754dd58f7"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.32,
+      "log_step": 168,
+      "text_zh": "观察：已逐帧检查视频 0.00–0.76 s 的全部20帧及两视角放大图；视频 0.12–0.40 s，MyoHand手指在红色圆柱上后方屈曲合拢，右侧视角可见手与圆柱间的空隙，未包夹圆柱。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 168,
+      "text_zh": "观察：视频 0.56–0.76 s，绿色参考标记升起并与红色圆柱分离，红色圆柱仍直立留在桌面，空合的手在其上后方；log step 168 的服务返回第11回合 control_steps=19、done=True，reason为物体距参考或手腕超过0.25米，评分器 sim_steps=19、live_matches_replay=1。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 168,
+      "text_zh": "推断：本次评分回合未拿起圆柱，也未完成要求的86个控制步及整段循迹，故本节点从未达成；初始圆柱与参考接近不代表整个循迹事件已完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "候选“空合手”：视频 0.12–0.40 s，单只MyoHand的拇指和其余手指在圆柱上后方合拢，未包围或夹住红色圆柱；这是已启动的抓取尝试，但没有可确认的目标接触，现有接触方式不能准确表示空合手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.32,
+   "log_step": 168,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.20–0.40 s，MyoHand在圆柱上后方空合，圆柱留在桌面；视频 0.56–0.76 s 参考标记升起而圆柱未跟随，该抓取失败直至片尾未修复。log step 168 确认这是第11回合，并由服务在19个控制步后报告物体距参考或手腕超过0.25米而终止。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.32,
+    "log_step": 168,
+    "evidence_zh": "观察：视频 0.00–0.76 s 的两个视角中，MyoHand由张开转为空合，右侧视角显示手指始终位于圆柱上后方，没有将圆柱夹入掌指之间；圆柱始终留在桌面，没有先被举起再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "log_step": 168,
+    "t": 0.32,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 168 的实际执行脚本按固定控制步调节腕伸肌和屈指肌；仅在循环索引k>=9且视觉物体估计非空时设置carry model offset并开始按持物模型循迹，没有要求圆柱随手运动或夹持成立。视频 0.12–0.48 s 显示手指在圆柱外侧上方闭合后离开，圆柱未被包夹，说明此次接近与合手的几何配合没有建立所需抓持；可行替代抓法未获验证。"
+   }
+  ],
+  "gap_zh": "无状态日志或接触对记录，不能核实瞬时轻触及接触力；右侧视角足以确认未形成包夹。评分器未报告track_pos_err、track_ok_fraction、track_max，故三项原生指标均为null，log step 168中agent脚本根据红色像素计算的err不能代替评分器精确质心误差。19个控制步按50 Hz对应仿真0.38 s，而视频时长为0.80 s，未建立可靠逐步映射，所有时间锚均直接取视频帧。agent在log step 177声称执行器能力使任务不可达，但其试验未排除所有合法策略，故不据此归因为任务或环境问题。"
+ },
+ "summary_zh": "评分的第11回合中，MyoHand在红色圆柱上后方空合，未拿起圆柱，圆柱循迹交接节点未达成，最终为0/1。视频 0.56–0.76 s 显示参考标记升起而圆柱留在桌面；log step 168 的服务在19个控制步后报告距离过大终止，固定合手及未经夹持确认便切换持物模型支持抓取和接触设计问题。无状态接触日志及评分器精确循迹指标，agent所称任务不可达尚未证实。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点（操作中）、没建立接触与抓取和接触设计的归因均与视频帧一致；时间锚取视频时钟，未发现需改动之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:36:02Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--fynmua5/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--fynmua5/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "4ff5f21146fa2ed2140f3382b339d42bdf4ffbe1f946061d558dfd9970f6ad44",
+   "video_sha256": "26042e16438c2a3c8c235563f9af817290f655424eef4d757dd28bf754dd58f7",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--fyuy5oe.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--fyuy5oe.json
new file mode 100644
index 0000000000000000000000000000000000000000..612bc25038c934115de38247e02a614f931fb257
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--fyuy5oe.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-standard--fyuy5oe",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 35,
+   "calls": 23,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 35,
+   "elapsed_s": 2247.5,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.499,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.266,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-standard--fyuy5oe/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "ac8aa5236c79ebe0edde95a0d1b46ba2d5e40c703e9d031346193eb84886187a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.24 s 的两个视角中，单只灵巧手向红色圆柱靠近并屈指，圆柱仍直立在桌面；尚未出现随手离开桌面的运动。"
+     },
+     {
+      "t": 0.56,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.28–0.76 s，手指在红色圆柱上缘附近蹭压，未见从两侧夹住圆柱；视频 0.56 s 后绿色参考标记升起并远离，红色圆柱仍在桌面，末帧视频 0.76 s 手指已与圆柱分离，未完成参考的升起、前送及回落。"
+     },
+     {
+      "t": null,
+      "log_step": 63,
+      "text_zh": "观察：log step 60 启动第2回合控制器；log step 63 读取该回合的 ep2_step_12 和 ep2_end 图像并打印 k=0–19 的关节诊断。评分器字段 episodes=2、resets=1、n_actions=19、sim_steps=19、task_failure=1、live_matches_replay=1，与这段短暂抓取失败的回放一致；没有完成节点要求的86个控制步。第1回合的辨识测试不计作本次进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频约 0.28–0.60 s，单只MyoHand的指端在红色圆柱上缘蹭压，随后抬离；两个视角均未显示拇指与其余手指从相对侧夹持承载圆柱，故实际接触记压/擦。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.56,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.56 s，灵巧手已对圆柱做屈指抓取动作，绿色参考标记开始明显上升，红色圆柱却仍立于桌面；至视频 0.76 s，手指空手抬离，圆柱未跟随。本回合在圆柱循迹交接的抓取起始阶段失败，此后没有修复。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.56,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.28–0.76 s，机器人指端虽蹭到红色圆柱上缘，但没有形成对向夹持和承载抓握；手指向上撤离时圆柱留在桌面。这里缺少的是可带动圆柱的抓握接触，未见先持起再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "log_step": 64,
+    "evidence_zh": "推断：log step 58 的控制脚本以保持初始腕姿并闭合拇指、食指和中指为目标；log step 64 的第2回合诊断却显示腕关节从 q_w=[0.740,0.071,-0.222] 漂移到 k=12 的 [0.572,0.285,-0.400]。结合视频 0.28–0.76 s 指端仅在圆柱上缘接触后空手抬离，控制器未能在执行中形成承载抓握；未验证接触布局本身是否可行，故仅归到控制，不细分反馈控制或抓取和接触设计。"
+   }
+  ],
+  "gap_zh": "观察：inputs/没有状态日志或成功示范；评分器reward未报告track_pos_err、track_ok_fraction、track_max，三项原生指标均保留null，不能从视频重建精确误差。log step 58 的脚本中段在输入日志内已被截断，第2回合结束的服务原始reason也未保留，无法精确确定原生终止条件或进一步细分控制根因。视频长0.8 s，而19个控制步按每步0.02 s仅为仿真0.38 s；未建立精确时钟映射，所标时刻全部直接取自已检查的视频帧，不把诊断的k值当视频时间。"
+ },
+ "summary_zh": "最终受评的第2回合中，灵巧手靠近并蹭压圆柱上缘，但未形成承载抓握；圆柱未随手抬起，圆柱循迹交接节点全程未达成，最终0/1。断点约在视频 0.56 s 的操作中，现象为没建立有效抓握接触，日志中的腕姿偏移支持控制层面的归因。没有状态日志和成功示范，三项精确跟踪指标未报告，控制子类及具体原生终止原因不能确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频仅0.8 s，逐帧所见与标签一致：指端蹭压圆柱上缘、未夹持，绿色参考上升而红色圆柱留在桌面；节点未达成、断点、接触、现象均有证据，根因标为推断且保守，validator通过。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:41:14Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--fyuy5oe/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--fyuy5oe/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "72e9fb1a69bbdd4afa351e4414f6f8f3d595fa613c87ed8d92c2ba143d243520",
+   "video_sha256": "ac8aa5236c79ebe0edde95a0d1b46ba2d5e40c703e9d031346193eb84886187a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--jdmgosg.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--jdmgosg.json
new file mode 100644
index 0000000000000000000000000000000000000000..3efd510c0e288ed1463bacf93d026801061e3090
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--jdmgosg.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-standard--jdmgosg",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 176,
+   "calls": 35,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 176,
+   "elapsed_s": 2418.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.362,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.372,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-standard--jdmgosg/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "f41cf772a5b5ddc016b238335a127dbac0a239a8d8193bc9e7064aa89430e473"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.4,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.76 s 的全部20帧；单只灵巧手的手指逐渐屈曲，在视频 0.40 s 已合拢于红色圆柱旁的空处。红色圆柱始终直立在桌面，未进入拇指与其余手指之间，也未随手离开桌面；视频 0.56–0.76 s 的绿色标记上移时，圆柱仍留在桌面。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 47,
+      "text_zh": "观察：log step 47 中机器人服务报告最终第11回合 control_steps=19、done=True，reason 为物体距参考或手腕超过0.25米而失败；评分字段 graded_by_live=1、n_actions=19、live_success=0。log step 50 再次查询仍为同一结束状态，故未完成节点要求的86个控制步。这里不把控制步或agent时间换算成视频时间。"
+     },
+     {
+      "t": null,
+      "log_step": 49,
+      "text_zh": "观察：log step 49 中agent在最终回合结束后才写出含执行器符号修正的脚本；log step 50 仅编译检查并查询已结束的服务状态，没有再次执行控制。该脚本的预期效果不属于视频中的评分尝试。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频 0.00–0.76 s，单只MyoHand尝试接近红色圆柱并屈指闭合，但圆柱始终在手外；候选方式“空合手（抓取未接触）”。没有形成对圆柱的包围或对向夹持，也未见推、压、托，因此现有实际接触方式均不适用，不记为已发生抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better"
+  },
+  {
+   "name": "track_ok_fraction",
+   "value": null,
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better"
+  },
+  {
+   "name": "track_max",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.4,
+   "log_step": 47,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.40 s 灵巧手已屈指合拢却未夹到红色圆柱，随后直至视频 0.76 s 圆柱仍留在桌面，循迹所需的拿持与带起没有实现。log step 47 的最终第11回合仅执行19个控制步即被机器人服务以物体距参考或手腕超过0.25米终止，未修复此次抓取失败。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.4,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.76 s 的两个并排视角均显示手指在红色圆柱旁闭合；视频 0.40、0.56、0.76 s 手与圆柱之间仍有间隔，圆柱未被包围、夹紧或带起。整个评分视频未出现先持住再滑脱的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent的接近与闭指控制未使实际手部到达并夹住圆柱。log step 46–47 的最终控制脚本用逆解输出直接填入前臂执行器命令并分阶段激活手指，log step 48 读出的第11回合本体感觉显示控制步0至14的ARTy由-0.126变为-0.135、ARTz约停在-0.100；结合视频 0.00–0.76 s 手部空合拢且圆柱未动，支持控制执行未实现抓取。agent在log step 49、51将其解释为执行器符号反向，但修正后未执行验证，不能据此确定更细控制子类。"
+   }
+  ],
+  "gap_zh": "输入中没有状态日志和成功示范，无法恢复精确接触对或逐步质心误差；run.json的评分字段未报告track_pos_err、track_ok_fraction、track_max，三项原生指标均保留null。19个控制步按任务控制周期仅为仿真0.38 s，而视频长0.80 s，没有可靠的逐帧控制步映射，因此所有视频锚点直接取自所看帧。执行器符号、肌肉耦合与伺服滞后对最终失败的各自贡献尚不能独立核实，根因仅标控制。"
+ },
+ "summary_zh": "最终评分的第11回合中，灵巧手接近并屈指，却始终未夹住或带起桌上的红色圆柱，圆柱循迹交接节点未达成，最终进度0/1。视频 0.40 s 可见空合手，log step 47 报告机器人服务在19个控制步后因物体距参考或手腕过远而终止；接近与抓取控制未实现所需运动，回合结束后写出的修正版未执行。没有状态日志、成功示范或原生跟踪指标，执行器反向等更细诊断尚未独立验证。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点（操作中）、没建立接触和停在“控制”的根因均与视频帧和log step 47/48/50的服务读数一致；视频时间均取自所见帧，未换算agent时间。未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:40:30Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--jdmgosg/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--jdmgosg/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "164627207a5b0a2f5e13c5eaab38164546466487bfeed67b98d1443a8a8091a5",
+   "video_sha256": "f41cf772a5b5ddc016b238335a127dbac0a239a8d8193bc9e7064aa89430e473",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--ltlnwqg.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--ltlnwqg.json
new file mode 100644
index 0000000000000000000000000000000000000000..39529d3e5a0c28aae00b80dcd60670fc7a04ed4a
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--ltlnwqg.json
@@ -0,0 +1,199 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-standard--ltlnwqg",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 160,
+   "calls": 88,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 160,
+   "elapsed_s": 858.4,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-standard--ltlnwqg/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "15fb84a4eda4ec3341de0b4c22fb2cc75b74c97af1048d2d1ec79fbfaec27c96"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.76 s 的两个并排视角，灵巧手在红色小圆柱旁收拢；视频 0.24 s 已可见手指收拢区域位于圆柱上方，圆柱没有进入手内。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 164,
+      "text_zh": "观察：视频 0.40–0.76 s，灵巧手继续抬高而红色圆柱仍直立在桌面；log step 164 显示最终脚本重置至第11回合、提交86行控制后只执行19个控制步，即以圆柱距参考或手腕超过0.25米的原因终止；log step 169 确认此后仍是同一终止回合。"
+     },
+     {
+      "t": null,
+      "log_step": 164,
+      "text_zh": "推断：最终回合未完成86个控制步，且视频没有圆柱随手升起、前送及回落的过程，因此整段循迹事件从未达成；不以初始位移误差接近零计作完成。视频时间直接取自帧，不把19个控制步换算成视频时刻。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–0.24 s，单只MyoHand灵巧手的手指在红色圆柱旁运动，随后在圆柱上方收拢并离开；未见拇指与其他手指从相对两侧夹住圆柱。两视角的初始局部投影重叠，无法确认是否曾短暂擦碰圆柱，故实际接触方式记未知，不将收拢动作记为已抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.24,
+   "log_step": 164,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.24 s，灵巧手已在圆柱上方收拢，圆柱仍在桌面；视频 0.40–0.76 s 手与圆柱进一步分开，未恢复抓取。log step 164 的最终第11回合执行19个控制步便报告圆柱距参考或手腕超过0.25米而终止，未完成整段循迹。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.24,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.08–0.40 s，灵巧手手指向内收拢，但红色圆柱始终处于收拢区域之外；视频 0.40–0.76 s 手空着抬起，圆柱继续留在桌面。这里缺失的是搬运所需的夹持接触，未见先握持再脱落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.24,
+    "log_step": 162,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 162 的最终控制脚本以基座俯仰减少0.15弧度作为抓取姿态，按固定时序先张指、再施加屈指及拇指肌肉激励，然后直接执行预计算的抬升轨迹；log step 164 确认该脚本用于最终回合。视频 0.16–0.40 s 显示实际收拢位置在圆柱上方、圆柱未被包夹，说明这组抓取姿态与手指闭合配置未建立承载接触，随后抬臂无法带动物体。没有验证成功的替代握法。"
+   }
+  ],
+  "gap_zh": "没有状态日志及成功参考视频，不能核实具体指节接触对、短暂擦碰或接触力；实际接触记未知。run.json 的评分字段未报告 track_pos_err、track_ok_fraction、track_max，三个原生指标均留 null。视频时长0.8秒与最终19个控制步乘0.02秒不相等，未建立控制步到视频的映射，所有视频锚点仅取自已检查的20帧。",
+  "unmet": {
+   "cylinder_reference_followed": false
+  }
+ },
+ "summary_zh": "最终评分第11回合中，灵巧手在圆柱旁收拢并空手抬高，圆柱始终留在桌面，圆柱循迹交接未达成（0/1）。log step 164 记录该回合仅执行19/86个控制步即因圆柱距参考或手腕过远而终止；视频支持所设抓取姿态及闭合配置未建立夹持这一控制原因。缺少状态接触对和原生跟踪误差统计，不能确定短暂擦碰或量化各项误差。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频、log 一致：最终第11回合手在圆柱上方空握抬起，圆柱留在桌面，节点未达成，断点与现象有证据支持。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:41:39Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--ltlnwqg/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--ltlnwqg/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "96eacab1bd0a00fbccc20e56758e92e2e96273c13c490514fec30ea1013d9d40",
+   "video_sha256": "15fb84a4eda4ec3341de0b4c22fb2cc75b74c97af1048d2d1ec79fbfaec27c96",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--njtyrjc.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--njtyrjc.json
new file mode 100644
index 0000000000000000000000000000000000000000..c7617ad85e3e97d569756eb7b20d498cd80a897d
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--njtyrjc.json
@@ -0,0 +1,198 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-standard--njtyrjc",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 215,
+   "calls": 180,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 215,
+   "elapsed_s": 2198.5,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-standard--njtyrjc/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "465177626d37520b70c3385e31fa53fe5d5ce48a7548f530c216517f8c02bfaf"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.28,
+      "log_step": null,
+      "text_zh": "观察：已检查视频0.00–0.76 s的全部20帧，两视角中MyoHand手指靠近并碰动红色圆柱，圆柱在桌面附近小幅移位；未见手指形成相向夹持并带圆柱离桌。"
+     },
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：视频0.60–0.76 s，灵巧手手指向上离开红色圆柱，圆柱仍在桌面；未完成随手抬升、前送及回落的整段运动。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 85,
+      "text_zh": "观察：log step 80执行最后一次reset，log step 81、83和85的服务状态均为episode=11、resets_left=0；log step 85中control_steps=19时done=true，reason为物体距参考或手腕超过0.25 m。评分器记录n_actions=19、live_matches_replay=1、task_failure=1；该评分回合没有完成要求的86个控制步。此处视频0.76 s仅为末帧导航锚点，不是把控制步或agent时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频约0.16–0.56 s，单只MyoHand的手指接触红色圆柱侧缘并将其在桌面附近拨移；视频0.60–0.76 s手指抬离，圆柱没有随手抬起。未见合手夹住圆柱，故记录实际拨动接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.6,
+   "log_step": 85,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约0.60 s起MyoHand手指抬离圆柱，至视频0.76 s末帧红色圆柱仍在桌面，未被带起完成循迹；log step 85的最终实时回合在control_steps=19报告物体距参考或手腕超过0.25 m并终止。断点位于正在执行的圆柱循迹交接步骤，视频时间直接取自逐帧检查。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.6,
+    "log_step": null,
+    "evidence_zh": "推断：视频0.00–0.56 s只见MyoHand手指靠近、碰动圆柱，未见相向夹持并带物体离桌；视频0.60–0.76 s手指抬开而红色圆柱留在桌面，说明用于抬升循迹的有效抓握没有建立。log step 80所说上一回合“grip slipped”是会话测试自述，不能用于判定本评分回合曾经抓住。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 80,
+    "t": 0.6,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 80的最终控制器按控制步推进合指及抬升，从control step 12起直接将拇指与食指尖中点作为已持圆柱的代理位置；log step 83修改手指目标和增益后继续同一回合。视频0.60–0.76 s显示手已抬开而圆柱仍在桌面，控制执行没有建立并带起物体。现有证据支持抓取与抬升控制失败，但不足以把接触几何、肌肉驱动跟踪或具体参数确定为唯一子类。"
+   }
+  ],
+  "gap_zh": "观察：inputs/中没有状态日志或成功参考视频；run.json评分字段未报告track_pos_err、track_ok_fraction、track_max，三项原生指标均填null。视频可见拨动及抬手后圆柱留桌，但没有接触对或力数据，不能确定具体接触指和失效的控制子类。视频为20帧、0.8 s，最终服务回合为19个0.02 s控制步；没有建立二者的精确时钟映射，所有视频锚点均直接按已检帧取得。"
+ },
+ "summary_zh": "评分回合为最后一次重置后的第11回合，唯一的“圆柱循迹交接”节点未达成，终态0/1。MyoHand手指碰动圆柱但未形成可带物体离桌的抓握，视频约0.60 s开始抬手后圆柱仍留桌；log step 85记录该回合在第19个控制步因物体距参考或手腕过远而终止，根因标为控制。缺少状态接触数据及三项循迹指标，无法进一步确定控制子类或精确误差。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "评分回合（最后一次重置后的第11回合）在第19控制步被原生终止，圆柱未随手完成循迹，节点false、终态0/1与视频和log step 85一致；断点、接触（推/扫）和单一现象/根因均有证据且措辞谨慎，未发现需要改变内容的错误。视频中圆柱在约0.6 s后留在桌面而参考球升走，无法确定是否曾短暂夹住（没建立接触 vs 没保持住），原标签已在gap_zh中声明此不确定。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:37:05Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--njtyrjc/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--njtyrjc/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b3437224aee20cce2c988772bb1d5c9038de7fbe83996744a2a2d191ec9e64c5",
+   "video_sha256": "465177626d37520b70c3385e31fa53fe5d5ce48a7548f530c216517f8c02bfaf",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--pmnthcu.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--pmnthcu.json
new file mode 100644
index 0000000000000000000000000000000000000000..1697a796b7a043b1e8814366ca19f644c3519ef5
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--pmnthcu.json
@@ -0,0 +1,199 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-standard--pmnthcu",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 198,
+   "calls": 31,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 198,
+   "elapsed_s": 2630.9,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-standard--pmnthcu/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "8dee38189d83e05d5a463b29605706ab62848e2561210cd1fbd5a7c79532abb2"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的两个外部视角均显示红色小圆柱立在桌面上，MyoHand手指在其旁边；初始静止不代表已完成整段参考循迹。"
+     },
+     {
+      "t": 0.48,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.76 s；视频 0.40–0.60 s，MyoHand手指靠近并碰动圆柱，圆柱向手外侧移动、倾斜，手指随后在圆柱旁卷曲；视频 0.64–0.76 s，手与圆柱分开，圆柱留在桌面附近，未随手完成升起、前送和回落。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 108,
+      "text_zh": "观察：log step 108 的机器人服务输出确认最后一次reset进入episode 11，执行11步后再执行8步，即control_steps=19时done=True，原因是圆柱距参考或手腕超过0.25米；log step 110再次确认同一终态。评分器报告n_actions=19、live_success=0、replay_success=0、live_matches_replay=1，未完成DAG要求的86个控制步。视频时刻直接取自所看帧，未用agent时间或控制步乘0.02秒换算。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.40–0.60 s，单只MyoHand的指端碰动红色圆柱，使其向手外侧移并倾斜；未见拇指与其余手指包合夹住圆柱，视频 0.64–0.76 s两者分开。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.48,
+   "log_step": 108,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.40–0.60 s，MyoHand尝试合指带动圆柱时，圆柱向手外侧移而未被包合；视频 0.64–0.76 s，手已离开，圆柱仍在桌面附近。log step 108 的服务输出显示评分回合episode 11在控制步19/86因圆柱距参考或手腕超过0.25米而终止，此前的抓取失败没有修复。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.48,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.36–0.60 s，MyoHand碰动红色圆柱后，手指在其旁边卷曲，圆柱向外侧移并倾斜；视频 0.64–0.76 s圆柱与手分开，未见手指包合圆柱并带其离开桌面。所缺的是搬运所需的抓握接触，实际发生过指端推碰。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.48,
+    "log_step": 107,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 107中agent为最终回合编写的脚本在控制索引9–12激活屈指和拇指对掌肌，同时从控制索引11开始把前臂ARRx目标由-0.80改为+1.75，试图在合指尚在进行时带起圆柱。视频 0.40–0.60 s显示手指没有包合圆柱，圆柱被推向手外；视频 0.64–0.76 s手离开而圆柱未被带走，说明这次合指与抬臂组合未建立可承载的接触构型。上述控制索引只描述脚本，视频时刻来自帧检查；没有验证可行替代握法，不据此认定任务不可解。"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功参考视频，不能精确核实接触指、接触力及是否出现极短的夹持；两外部视角足以确认未建立能带走圆柱的抓握。评分器reward未报告track_pos_err、track_ok_fraction和track_max，三项值均为null，不能从图像重建厘米级误差或从agent自述证明任务不可解。服务只提供终止原因文字，未单独提供原生info.done原值；视频锚点均直接取自25 fps帧，不与日志控制步时钟换算。"
+ },
+ "summary_zh": "评分回合为最后的episode 11：MyoHand尝试合指抓取，但仅将圆柱推向外侧并使其倾斜，未建立能带起圆柱的抓握，圆柱循迹交接节点未达成（0/1）。失败断在该步骤操作中，接触构型与抬臂配合未能承载圆柱；log step 108确认服务在控制步19/86提前终止。无状态日志、成功参考视频或逐步误差统计，无法量化接触力及三项原生误差指标。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点在唯一步骤操作中、接触为指端推碰、控制步19终止均与视频帧和log step 108一致；视频时间未由agent时间换算。未发现需改内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:36:29Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--pmnthcu/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--pmnthcu/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a513974fd0b2e587a647669b4ddf5bbbd420e58940c47a024f6ad929ae2a1d8c",
+   "video_sha256": "8dee38189d83e05d5a463b29605706ab62848e2561210cd1fbd5a7c79532abb2",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--tjsaocx.json b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--tjsaocx.json
new file mode 100644
index 0000000000000000000000000000000000000000..584ffa38e7756904ace004bc6717106795c99fef
--- /dev/null
+++ b/phase2/myosuite/dex_cylinder_pass/trajectories/myosuite-dex-cylinder-pass-i00-standard--tjsaocx.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-cylinder-pass-i00-standard--tjsaocx",
+ "taskKey": "myosuite/dex_cylinder_pass",
+ "dag_sha256": "f6817fb6c38513c6f5912bbbb253639b3a62683d2ab63a20e565756a17e50355",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 19,
+   "sim_steps": 19,
+   "physics_steps": 190,
+   "steps": 128,
+   "calls": 47,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 134,
+   "elapsed_s": 709.6,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.57,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.835,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-cylinder-pass-i00-standard--tjsaocx/replay.mp4",
+  "duration_s": 0.8,
+  "fps": 25.0,
+  "frames": 20,
+  "sha256": "8e19c2beb8eed333ebcf84b71e4326c95ba1bb5c129f0c7941b4f821c4a9af25"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cylinder_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.32,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.76 s 的全部20帧及手部放大图；MyoHand手指逐渐屈曲并远离圆柱，红色圆柱始终留在桌面附近，未见被夹住后随手抬起。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 44,
+      "text_zh": "观察：log step 44 的服务响应为 episode 11、resets_left 0、control_steps 0；其后 log step 45–49 连续提交动作且无重置。评分器 reward 记录最终回合 n_actions=19、sim_steps=19、task_failure=1；视频末帧圆柱仍在桌面附近。"
+     },
+     {
+      "t": 0.76,
+      "log_step": 52,
+      "text_zh": "推断：最终评分回合仅执行19个控制步，且视频未显示圆柱被抬起，未完成要求的全部86步循迹，故该事件节点从未达成。agent 自述最终回合在控制步19因圆柱距参考或手腕超过0.25米而结束（log step 52）；此自述仅作补充，不将agent时钟换算为视频时钟。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频 0.00–0.76 s，单只MyoHand手指在圆柱旁屈曲，但未见手指围住或对向夹紧圆柱，圆柱没有被带起。候选方式“空抓（未触物的抓取尝试）”：现有方式没有未建立目标接触的取值，不将空抓标为已发生抓握；无状态日志可核实帧间短暂轻触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok_fraction",
+   "unit": "1",
+   "threshold": 0.9,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cylinder_reference_followed",
+   "t": 0.32,
+   "log_step": 48,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.16–0.44 s，MyoHand手指屈曲并向上收拢，圆柱留在手指外的桌面上；至视频 0.76 s 仍未拿起。log step 45–49 的动作代码确实在最后回合调整前臂并闭指，评分器记录该回合仅19个控制步，未完成循迹。"
+  },
+  "unmet": {
+   "cylinder_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.32,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.76 s 两个视角及逐帧放大图均未见MyoHand对向夹住圆柱；视频 0.24–0.44 s 手指已收拢到圆柱上方、侧方，圆柱仍留在桌面附近，之后也未随手升起。没有先持住再滑落的可见过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：最后回合中，agent脚本先以固定前臂目标及轻度闭指命令尝试抓取（log step 45），再改变前臂目标并增强手指肌肉命令（log step 46–49）；视频 0.16–0.76 s 显示实际手指运动没有形成对圆柱的夹持。控制尝试未把手指闭合在物体上，是循迹无法开始的直接原因；现有证据不能进一步区分接触几何、肌肉协调或前臂跟踪问题。"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功参考视频；最终回合多次动作调用的out.text为空，0.25米终止说明仅见agent在log step 52的自述，评分器只确认task_failure。视频足以判断未夹持、未抬起，但不能核实帧间轻触、精确接触力及控制子类；评分器未报告track_pos_err、track_ok_fraction、track_max，故三项原生指标均为null。视频时刻仅按帧定位，未将19个控制步或agent时间换算为视频时间。"
+ },
+ "summary_zh": "最后第11回合中，MyoHand尝试调整前臂并闭指，但圆柱未被夹持或抬起，“圆柱循迹交接”未达成，最终为0/1。断点在该步骤的操作中，现象为没建立接触，归因为控制层面的抓取执行失败；评分器记录最终回合仅19个控制步。无状态日志和原生循迹误差数值，无法细分控制原因；log step 50之后保存但未完整运行的控制脚本不作为评分轨迹。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、没建立接触、根因停在“控制”均与视频一致：参考绿球升起远离，红色圆柱始终留在桌面、未被夹持；未把agent时间换算成视频时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:38:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--tjsaocx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-cylinder-pass-i00-standard--tjsaocx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "254cf674ae6226df37fa204014390b6f4918de4fd0d7f494e937f81bc422c2fe",
+   "video_sha256": "8e19c2beb8eed333ebcf84b71e4326c95ba1bb5c129f0c7941b4f821c4a9af25",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--3tyvler.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--3tyvler.json
new file mode 100644
index 0000000000000000000000000000000000000000..f9458adeb866ca3c0de42790b6437724e0017d5a
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--3tyvler.json
@@ -0,0 +1,191 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-privileged--3tyvler",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 99,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 77,
+   "agent_resets": 130,
+   "agent_sim_steps": 6693,
+   "agent_records": 22,
+   "agent_replays": 21,
+   "agent_cost_self_reported": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-privileged--3tyvler/replay.mp4",
+  "duration_s": 4.0,
+  "fps": 25.0,
+  "frames": 100,
+  "sha256": "706217c34e44fd105c650c2ab40c63b54829f80362a4850719f3b811d4758974"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.80 s，MyoHand 单手的手指围在棕色锤柄尾端，手和锤柄共同移动；视频 1.20–1.60 s，锤子离开原来的桌面位置并转向，锤柄仍夹在手指之间。该步骤已实际尝试。"
+     },
+     {
+      "t": 1.2,
+      "log_step": 128,
+      "text_zh": "推断：视频 0.80–1.60 s 的抬升、转向阶段，锤子与画面中的绿色球形参考标记明显分离；结合 log step 128 对交付文件的两次新进程重放结果（均执行 99 步，track_ok=40，最大位移误差约 0.187 m，success=false，终态哈希均为 8cde2753231bcd8b），整段参考运动未通过，不能因末段接近参考而记为达成。视频锚点来自所看帧，未将 agent 工作时间换算为视频时间。"
+     },
+     {
+      "t": 3.96,
+      "log_step": 126,
+      "text_zh": "观察：视频 3.00–3.96 s 逐帧可见手仍围住锤柄，锤子随手小幅运动，没有掉回原位置。log step 126 的 late 候选完成 99 步，末步 track_pos_err=0.00997 m、track_rot_err=0.0138 rad、track_ok=40、原生 info.done=false；log step 127 明确将 late 的 99×45 条执行器命令导出为交付文件，log step 128 重放终态哈希与该候选一致。末步误差达标不补回过程中缺失的合格步数或历史最大误差。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.80–1.60 s 及视频 3.00–3.96 s，MyoHand 单手的拇指侧和其余弯曲手指从锤柄两侧围夹，带动锤子抬升、转向；锤子自身是受评物体。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": 0.00997
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": 0.0138
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": 40
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": 0.187
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 1.0,
+   "log_step": 128,
+   "where": "操作中",
+   "evidence_zh": "推断：视频约 0.80–1.40 s，MyoHand 正握持锤柄进行抬升和转向，锤子与绿色球形参考标记在画面中明显分离；log step 127–128 证明交付的就是 late 轨迹，并以一致的两次重放确认仅 40/99 步合格、最大位移误差约 0.187 m。整段跟踪步骤最终未修复；视频约 1.00 s 只作中段偏离的导航锚点，不声称是首次越过数值阈值的精确时刻。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.2,
+    "log_step": 128,
+    "evidence_zh": "推断：视频 0.80–1.60 s，手带锤子抬升、翻转时，锤子所在位置与绿色球形参考标记明显错开，锤头随后才转至前臂旁；视频 3.00–3.96 s 锤柄仍在手中。结合 log step 126–128 对同一交付轨迹的误差和重放记录，失败表现为挥锤过程的位置、朝向跟踪偏离，而末步已经回到误差容限内。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 122,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 114 记载 agent 自述当前完整候选的抬升滞后、旋转不对，并以它继续优化；log step 122 的脚本从 wrist45 的前缀继续搜索执行器命令，log step 126 输出得到的 late 轨迹末步可对准、过程仍大幅偏离。log step 127 导出该动作序列，log step 128 两次重放均复现相同失败与终态，视频 0.80–1.60 s 也显示手确实持锤转向。综合支持执行器运动控制未能实现所需过程跟踪；没有足够证据区分具体握持几何、轨迹求解和反馈控制子类。"
+   }
+  ],
+  "gap_zh": "没有逐帧仿真状态日志、完整逐步误差序列或成功 oracle；视频可以核对握持和抬升转向，但不能独立测出厘米级误差或首次越界时刻。native_metrics 的末步误差与 track_ok 来自 log step 126 的 benchmark Session 状态，并由 log step 127–128 的导出与相同终态重放关联到交付轨迹；track_max=0.187 m 仅为 log step 128 重放 reason 中报告的三位小数近似值，reward 汇总字段未提供更高精度。接触力和具体控制失配机制未记录，因此根因不细分。"
+ },
+ "summary_zh": "MyoHand 单手抓握锤柄并带动锤子抬升、转向，交付轨迹完整执行 99 步，但仅 40 步跟踪合格、最大位移误差约 0.187 m，沿参考挥锤节点最终为 0/1。末步虽已对准，中途的位姿偏离使整段验收失败；两次一致重放及视频支持控制层面的过程跟踪失败。缺少逐步误差和接触力记录，无法精确确定首次越界时刻或进一步细分控制根因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点、位姿不准现象与控制根因均有视频和日志(step 126-128)支持，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:04:47Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--3tyvler/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--3tyvler/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1305edb6eb444fb0df5bf3dce22a03eef0dbcd6d6df69088a410dbd947aef3b1",
+   "video_sha256": "706217c34e44fd105c650c2ab40c63b54829f80362a4850719f3b811d4758974",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--cugbt6g.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--cugbt6g.json
new file mode 100644
index 0000000000000000000000000000000000000000..ab92a369e3529b28aff5419c784b016e5f19bcd9
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--cugbt6g.json
@@ -0,0 +1,205 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-privileged--cugbt6g",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 54,
+   "sim_steps": 54,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 78,
+   "agent_resets": 232,
+   "agent_sim_steps": 8313,
+   "agent_records": 112,
+   "agent_replays": 15,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.054,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.244,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-privileged--cugbt6g/replay.mp4",
+  "duration_s": 2.2,
+  "fps": 25.0,
+  "frames": 55,
+  "sha256": "c915211a5e308b3d0bea0dea7d2b331846cee51349e67dfdc4ed8c805133870a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–2.16 s 的两个并列视角；MyoHand 手指在锤柄上方屈伸并接近、触碰锤柄，锤子始终贴着桌面，未出现被手带离桌面及往复挥动；视频 0.80 s、1.20 s、2.16 s 可核对。"
+     },
+     {
+      "t": null,
+      "log_step": 50,
+      "text_zh": "观察：log step 49 的 agent 交付生成脚本记录并保存动作，log step 50 输出保存 54 个控制步，随后对该动作序列重放，success=False、track_ok=18；它未完成任务规定的 99 个控制步。"
+     },
+     {
+      "t": null,
+      "log_step": 87,
+      "text_zh": "观察：log step 54 对交付文件的两次独立重放均在 54 个控制步失败且终态哈希一致；log step 87 最后复核仍为 instance_ok=True、deterministic=True、success=False，原生终止原因为锤子距参考或腕部超过 0.25 m。此为交付重放结果，不采用后续会话中其他试验的 99 步结果。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.16 s 及约 0.76–1.24 s，MyoHand 单手的手指在桌上锤柄上方接触、触压；侧视未见手指从锤柄两侧形成承载包夹，锤子没有离桌。实际接触按触压标注，不照搬 DAG 的典型抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 0.8,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80 s，MyoHand 手指在锤柄上方活动，但锤子仍平放桌上；随后至视频 2.16 s 均未抬起挥动。log step 50、54、87 的交付重放确认只执行 54 个控制步便因锤子距参考或腕部过远终止，未完成沿参考挥锤。视频锚点取自帧，不将控制步或 agent 用时换算为视频时间。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "推断：视频 0.00–2.16 s 双视角显示 MyoHand 手指在锤柄上方屈伸、间歇触压，锤子一直贴着桌面，未随手抬起或摆动；没有形成这次抬锤所需的承载抓握。“没建立接触”指所需抓握未建立，并非完全没有皮肤与锤柄的触碰；未见已持锤后再掉落的阶段。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "视频 / 日志",
+    "evidence_zh": "推断：视频 0.00–0.16 s、0.76–1.24 s 的侧视中，MyoHand 手指接近、触压锤柄上方，未形成把锤柄包住并承载抬起的接触布局；锤子至视频 2.16 s 仍留在桌上。log step 49 的交付脚本以参考手部关节位姿作 PD 目标，经逆动力学及 NNLS 生成肌肉命令，未根据锤柄的实际接触状态重建抓握；结合视觉结果，判为该交付的抓取接触布局不适配。agent 在 log step 88 自述手指压在锤柄上而未绕到下方，与视频一致；该自述中关于初态偏差和其他握拳试验的具体机制不据此视为已证实。没有成功替代抓法证据。"
+   }
+  ],
+  "gap_zh": "无状态日志、接触对或接触力记录，不能确定各手指的实际法向力及接触瞬间；抓取几何导致抬锤失败为视频与交付控制代码支持的推断。无成功 oracle。run.json 的 reward 未报告 track_pos_err、track_rot_err、track_ok、track_max，故四项 native_metrics 值保留 null；log step 50 的交付自重放另报 track_ok=18，但没有交付的完整逐步误差或历史最大值。视频长 2.20 s，交付的 54 个控制步按任务控制周期对应仿真 1.08 s，未建立两钟映射，全部视频时间直接按帧号及 25 fps 取值。"
+ },
+ "summary_zh": "交付轨迹未完成唯一节点“沿参考挥锤”（终态 0/1）：视频 0.00–2.16 s 中手指在锤柄上方活动、触压，锤子始终留在桌面，没有被握持抬起挥动。log step 50、54、87 确认交付在 54 个控制步因锤子距参考或腕部过远终止；结合视频和交付控制代码，失败归于未建立承载抓握及抓取接触设计不适配，精确接触力、误差历史与可行替代握法因缺少相应记录而无法核实。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点未完成、最终0/1、断点与现象/根因均与视频和交付重放结果一致；未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:06:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--cugbt6g/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--cugbt6g/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "930327e67994279db299262a92c6e91ffee9c30d815e786c5799906b587785c8",
+   "video_sha256": "c915211a5e308b3d0bea0dea7d2b331846cee51349e67dfdc4ed8c805133870a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--fngk2wa.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--fngk2wa.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c129d4cefbfc191824c742001fec5c8bd0c0dbf
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--fngk2wa.json
@@ -0,0 +1,227 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-privileged--fngk2wa",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 18,
+   "sim_steps": 18,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 27,
+   "agent_resets": 72,
+   "agent_sim_steps": 303,
+   "agent_records": 15,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.556,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.979,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-privileged--fngk2wa/replay.mp4",
+  "duration_s": 0.76,
+  "fps": 25.0,
+  "frames": 19,
+  "sha256": "4b1d97a04b3140a6013856d5c99fc41414f82d08b7e6b057335d9595b777cd1c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 70,
+      "text_zh": "观察：视频 0.00–0.72 s 的全部 19 帧中，MyoHand 手指初期贴碰桌面上的锤柄，随后与锤子分离；锤子只在桌面滑移，没有被握起或往复挥动。log step 70 的最终录制代码将六个手臂执行器保持在初始目标，39 个肌肉命令均设为 -1，没有调用会话中试验过的跟踪控制器。"
+     },
+     {
+      "t": 0.72,
+      "log_step": 70,
+      "text_zh": "观察：log step 70 保存并重放最终 18×45 动作文件，基准 Session 返回 steps=18、track_steps=99、track_ok=18、success=False，以及物体距参考或腕部超过 0.25 m 的终止原因；inputs/run.json 的评分器字段 n_actions=18、sim_steps=18、deterministic=1、task_failure=1，与视频中锤子未离桌且短片提前结束相符。"
+     },
+     {
+      "t": 0.72,
+      "log_step": 70,
+      "text_zh": "推断：最终轨迹只有 18 个控制步，未完成节点要求的全部 99 步；初期 track_ok 的累计不代表整段参考运动完成，因此该节点从未成立。视频锚点按已查看的帧时间给出，不将 log step 70 的控制步数或 agent 时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 0.00–0.24 s，MyoHand 单手手指贴碰锤柄，锤子沿桌面滑移；随后手与锤柄分离。实际只有初始贴碰和滑动，没有形成拇指与其余手指的对向抓握；这些是基线中的被动接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 0.0,
+   "log_step": 70,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0.00–0.72 s，锤子始终在桌面，MyoHand 没有握起并挥动锤子；log step 70 最终交付代码只保持手臂初始目标并将肌肉命令设为 -1，实际录到 18 步后以原生距离失败终止。推断：交付的是未启动抓握及挥锤动作的放松基线，断在“沿参考挥锤”的首步前；会话里的 DLS 或握持试验不属于交付运动。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 70,
+    "evidence_zh": "观察：视频 0.00–0.72 s 未见主动合手抓握、抬锤或挥锤；log step 70 保存的动作始终令肌肉放松、手臂目标固定，且 agent 在 log step 71 明确称其为最终基线。推断：该评分轨迹从未启动所要求的参考挥锤动作，初始贴碰及被动滑移不构成挥锤尝试。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 58,
+    "source_zh": "日志",
+    "evidence_zh": "推断：会话中的控制试验未能建立带动锤子的握持，使 agent 最终只能交付放松基线。依据是 log step 58 的诊断输出：不同挤压设置下，脚本驱动腕部运动而锤子高度基本维持约 0.012 m；log step 62 的六组握持跟踪测试均在 16–17 个控制步触发距离终止；log step 70 最终录制放弃这些控制器。上述是会话层的控制能力证据，不作为评分视频里的动作或节点进度；现有记录不足以细分为接触几何、轨迹求解或反馈控制问题。"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功 oracle，视频也没有完整参考位姿叠加，无法核实逐步接触力、精确接触几何及原生终止的具体距离分支。评分器 reward 未报告 DAG 的四项原生指标，native_metrics 的 value 均留 null；log step 70 的最终自验重放虽报告位置误差 0.02489 m、旋转误差 0.1746 rad 和 track_ok=18，未将其冒充评分器 reward 指标。agent 在 log step 71 关于参考手姿不可行的判断不能排除所有合法控制方案，故不据此归因任务或环境。"
+ },
+ "summary_zh": "评分轨迹只执行了 18 个控制步，锤子始终留在桌面，唯一节点“沿参考挥锤”未达成，最终进度为 0/1。交付的放松基线未启动抓握与挥锤，断点位于首步前；会话试验显示控制器未能让锤子随腕部运动，支持一般“控制”根因。缺少状态日志、成功 oracle 及评分器原生误差指标，无法进一步确定控制子类。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "WHERE 标签（节点未达、断点首步前、没尝试）与视频和交付动作一致；问题在根因：“控制”完全建立在 agent 会话内的试验和自述上，不是评分局证据，应清空并写 gap_zh。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达、断点首步前、没尝试均与视频及交付动作一致；第1轮提出的根因争议，labeller 引用 03 文档和 FAIL-22（会话抓取失败可作为控制根因证据）已有依据，无新证据，不再重复。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:52Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--fngk2wa/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--fngk2wa/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--fngk2wa/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--fngk2wa/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "05fe6d7c2278bb1d364f07a294ebed27329233a8d7a5a8380534be002f8920ae",
+   "video_sha256": "4b1d97a04b3140a6013856d5c99fc41414f82d08b7e6b057335d9595b777cd1c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--gnvxcdb.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--gnvxcdb.json
new file mode 100644
index 0000000000000000000000000000000000000000..ed6626f67ffa489baac2bb2a34dd569ffe11ece4
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--gnvxcdb.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-privileged--gnvxcdb",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 99,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 114,
+   "agent_resets": 3492,
+   "agent_sim_steps": 156701,
+   "agent_records": 0,
+   "agent_replays": 23,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.189,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.841,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 1,
+   "session_log_from_monitor": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-privileged--gnvxcdb/replay.mp4",
+  "duration_s": 4.0,
+  "fps": 25.0,
+  "frames": 100,
+  "sha256": "f336ac00da76f61399821b5bbdf2f178f069e6669bd0bc48700069b410991ae4"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.64 s 的两个视角显示 MyoHand 的拇指与弯曲手指围在棕色锤柄两侧，锤子仍近于初始横向姿态；初始接触和初始位置接近不等于完成整段参考运动。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：视频 0.80–1.56 s 中手腕及锤柄逐渐转向，绿色标记从锤子附近移向其上方及侧方；视频 1.60–3.96 s 中锤子在手旁基本保持同一斜向姿态，绿色标记持续往复移动，未见相应的完整往复挥锤。"
+     },
+     {
+      "t": null,
+      "log_step": 184,
+      "text_zh": "观察：log step 180 将 cem_full_phase1 候选保存为交付文件，log step 183 运行其生成器后，log step 184 的两次重放均执行 99 行、success=false、终态哈希 fa726dcb18321b92，并报告仅 18/99 步在轨、末步偏离、最大位移误差约 0.256 m；与评分器字段 deterministic=1、sim_steps=99、task_failure=1 一致。"
+     },
+     {
+      "t": null,
+      "log_step": 123,
+      "text_zh": "推断：log step 123 中候选的终态哈希同为 fa726dcb18321b92，且 log step 180 明确交付该候选，log step 184 再次得到同一哈希及在轨计数，故采用该候选服务状态的末步 track_pos_err=0.14836 m、track_rot_err=1.3478 rad 和 track_ok=18；track_max=0.256 m 取交付重放判定文字的舍入值。整段验收从未成立，不能把开头的合格采样段计作节点完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.00 s、0.96 s 及 1.60–3.96 s，单只 MyoHand 的拇指与弯曲手指从两侧包围锤柄，锤柄位于合拢的手内；操作对象是锤子本身。未见明确脱手事件，抓握记录不代表参考轨迹已达标。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": 0.14836
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": 1.3478
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": 18
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": 0.256
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 0.96,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80–1.20 s，MyoHand 开始转动手腕，锤子仍停留在手旁较低的位置，绿色标记向上及侧方移开；视频 1.60–3.96 s 仍未出现锤子跟随标记往复的运动。log step 184 的交付重放确认整段仅 18/99 步在轨，末步及历史最大误差均未通过。视频 0.96 s 是可见偏离的导航锚点，不是由 agent 工作时间换算的精确阈值越界时刻。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.96,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.96–1.56 s 锤子转至斜向后，视频 1.60–3.96 s 锤柄和锤头基本保持该姿态；绿色标记则反复移向锤头外侧、画面下方再返回。锤子没有呈现与该移动标记相应的往复位移，且 log step 184 的交付重放报告末步仍偏离参考。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 120 的候选控制代码在开头保持前臂目标，随后施加搜索得到的前臂轨迹和固定肌肉组兴奋水平；log step 180 将该候选冻结为 99 行开环命令。log step 123 的服务输出和 log step 184 的交付重放均显示该命令序列只能维持 18 个在轨采样步，视频 0.96–3.96 s 也未实现要求的往复挥锤，支持执行器控制未能产生所需物体运动；现有证据不足以细分为接触设计、逆解或反馈控制问题。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志、接触力记录、完整参考姿态叠加或成功 oracle；不能精确定位首次厘米级误差越界，也不能确认握持承载能力或控制失败的具体子类。视频时长为 4.0 s，任务的 99 个控制步为仿真 1.98 s，本文仅用实际视频帧定位，不将两个时钟直接等同。原生指标来自已由候选选择及相同终态哈希关联的服务结果，最大误差只有重放文字中的三位小数。"
+ },
+ "summary_zh": "MyoHand 的手指围握锤柄并使锤子转向，但没有完成参考要求的往复挥动，唯一节点“沿参考挥锤”未达成，最终为 0/1。视频约 0.96 s 起可见运动偏离，交付文件两次确定性重放均仅 18/99 步达标，末步偏离且最大位移误差约 0.256 m，归为位姿不准及控制失败。缺少逐帧状态和接触力，无法确定具体控制子类或首次精确越界时刻。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点未达成、断点、位姿不准与控制根因均有视频（0.0/0.96/2.0/3.96 s）及 log step 123/180/184 的服务读数支持；未发现需要改动的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:09:41Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--gnvxcdb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--gnvxcdb/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "eaeb092d534814138fc836bd35f32560e9a296a65ed1ed9bb2c5ce834e4f9e19",
+   "video_sha256": "f336ac00da76f61399821b5bbdf2f178f069e6669bd0bc48700069b410991ae4",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--ididclt.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--ididclt.json
new file mode 100644
index 0000000000000000000000000000000000000000..f66448481c03a525e3cdaed75a8340a2cea0c375
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--ididclt.json
@@ -0,0 +1,217 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-privileged--ididclt",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 60,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 423,
+   "agent_resets": 435,
+   "agent_sim_steps": 8721,
+   "agent_records": 5,
+   "agent_replays": 14,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.276,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.276,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-privileged--ididclt/replay.mp4",
+  "duration_s": 2.44,
+  "fps": 25.0,
+  "frames": 61,
+  "sha256": "072ca735fb71db3b44f7b7e8d26f4bb2abda298e9bd886038619d10944e2ad9f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.64 s 的俯视与侧视中，MyoHand 手指贴在桌面上的锤柄上方；未见锤子被包夹提起。"
+     },
+     {
+      "t": 0.88,
+      "log_step": null,
+      "text_zh": "观察：视频 0.72–1.08 s，MyoHand 手指与腕部离开锤柄，锤子留在桌面，没有随手臂抬起；视频 1.40–2.40 s 手臂再次扫过锤子附近，锤子被碰移，但没有完成持锤往复挥动。"
+     },
+     {
+      "t": 2.4,
+      "log_step": 577,
+      "text_zh": "观察：log step 562 重新生成最终交付文件后，log step 563、577 的官方重放均报告 n_actions=99、replayed_rows=60、success=false、deterministic=true，终止原因为物体距参考位置或腕部超过 0.25 m；inputs/run.json 的 reward.sim_steps=60、task_failure=1 与其一致。"
+     },
+     {
+      "t": 2.4,
+      "log_step": 577,
+      "text_zh": "推断：视频中的抬手未带锤及官方重放仅执行 60 个控制步共同表明，整段 99 步参考运动的验收从未完成；开局姿态接近或短暂误差合格不构成该整体节点达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.72 s，MyoHand 单手的指腹贴压桌面上的锤柄，未见对向包夹后承载锤子的抓握。"
+     },
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 1.40–2.40 s，MyoHand 单手及腕部再次扫过锤柄附近，锤子沿桌面被拨动、改变朝向，未形成随手挥动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 0.8,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.80 s，MyoHand 已开始抬腕运动，但手指从锤柄上方离开，锤子仍留在桌面；视频 0.88–1.24 s 手与锤子明显分离，后续至视频 2.40 s 未完成持锤轨迹。log step 577 的最终交付重放在 60 个控制步后触发物体偏离过远的原生终止。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.72 s，MyoHand 手指接触锤柄上方；视频 0.80–1.08 s，手指抬起而锤子仍在桌面。未见能把锤子带离支承面的包夹握持，所需的承载抓握没有建立；这里有表面接触，并非手从未碰到锤子。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 0.8,
+    "log_step": 68,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：视频 0.00–1.08 s 中，MyoHand 的指腹主要从锤柄上方接触，抬腕时没有形成承载包夹。log step 68 对同一固定初态的会话诊断显示，拇指、食指及中指接触法向主要朝下，锤子同时由桌面支承；log step 559 的最终控制方案保留初始手指姿态并施加屈指偏置后执行抬腕。结合视频中手离开而锤子不随行，判断失败源于未建立合适的承载接触布局；未验证可行替代握法。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志或成功 oracle；视频可判定抬手未带起锤子及轨迹未完成，但不能精确重建指腹接触力或全部误差曲线。reward 未报告 track_pos_err、track_rot_err、track_ok、track_max，故四项原生指标均填 null；日志中的早期控制方案诊断属于会话证据，仅用于辅助根因，不当作最终交付轨迹的逐步指标或视频时刻。",
+  "unmet": {
+   "hammer_reference_followed": false
+  }
+ },
+ "summary_zh": "交付轨迹未完成唯一节点“沿参考挥锤”，最终进度为 0/1；视频约 0.80 s，手指抬离锤柄而锤子留在桌面，未建立能承载挥动的抓握。最终交付的两次官方重放均在 60 个控制步后因物体偏离过远而终止，未完成要求的 99 步。根因判断为抓取和接触设计不足，但缺少交付轨迹的接触力、完整误差记录及成功参考视频。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "唯一节点未达成、final=0、断点与现象与视频及官方重放结果一致；视频帧中手离开锤柄、锤子被扫移，两次重放均在 60 步因物体偏离而终止。接触与根因为推断且已标注，未发现需要修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:05:09Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--ididclt/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--ididclt/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "fb1233e2b2dc1c28cbe784fd863e75605296905c6b4efcbf546eadf8e21049d4",
+   "video_sha256": "072ca735fb71db3b44f7b7e8d26f4bb2abda298e9bd886038619d10944e2ad9f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--ugrcnag.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--ugrcnag.json
new file mode 100644
index 0000000000000000000000000000000000000000..1d2ab6e8189eabd0989fc18d0d42b1202e3666b2
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-privileged--ugrcnag.json
@@ -0,0 +1,205 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-privileged--ugrcnag",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 99,
+   "sim_steps": 99,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 117,
+   "agent_resets": 42382,
+   "agent_sim_steps": 8729,
+   "agent_records": 9,
+   "agent_replays": 10,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.674,
+   "peak_processes": 27,
+   "peak_disk_gb": 0.505,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-privileged--ugrcnag/replay.mp4",
+  "duration_s": 4.0,
+  "fps": 25.0,
+  "frames": 100,
+  "sha256": "a86825ddd05d4fb68c44ccc6e44b4a7c78ef09baa943ca7a736b40191874088c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.4,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.52 s，MyoHand 手指在锤柄末端弯曲，锤子仍接近桌面；视频 0.80–1.36 s，手带动锤子离开原处，锤头逐渐垂到手下方，实际开始了持锤运动。"
+     },
+     {
+      "t": 3.96,
+      "log_step": 91,
+      "text_zh": "观察：视频 1.20–3.96 s，锤柄末端仍连在弯曲手指处，锤头悬垂，后段锤柄由斜向转为近竖直。log step 90 生成并保存最终 trajectory.npz；log step 91 的官方 replay 命令对该文件在两个新进程重放，均报告执行 99 步、仅 21 步 on track、末步 off track、最大位移误差约 0.084 m，两个终态哈希相同；run.json 的 reward 同时记录 success=0、deterministic=1、sim_steps=99。"
+     },
+     {
+      "t": 3.96,
+      "log_step": 91,
+      "text_zh": "推断：最终交付文件完成了控制步数，但累计达标步数、末步和历史最大位移误差验收失败，因此整段「沿参考挥锤」从未成立，不能把初始接近参考或短暂带起锤子算作节点完成。视频时间均直接取帧时间，未将控制步或 agent 时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "勾",
+      "note_zh": "观察：视频 0.40、1.00、2.00、3.96 s，MyoHand 弯曲手指勾住锤柄末端并带动锤子，锤头悬在接触处下方；可见的是末端勾持，未见拇指与其他手指对向夹紧锤柄的明确证据。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": 21
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": 0.084
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 1.2,
+   "log_step": 91,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 1.20 s，MyoHand 已带起锤柄末端，但锤头向下悬垂；此后至视频 3.96 s 仍以这种末端持锤姿态移动。log step 91 对最终交付文件的两次官方重放均报告仅 21/99 步达标、末步未达标、最大位移误差约 0.084 m；失败发生在持续操纵锤子的步骤，视频 1.20 s 仅为可见姿态问题的导航锚点，不是推算的首次误差越界时刻。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.2,
+    "log_step": 91,
+    "evidence_zh": "观察：视频 1.20–3.96 s，锤子主要以柄尾贴住弯曲手指、锤头垂在下方的姿态移动，后段锤柄逐渐接近竖直；锤子没有从手中掉落。log step 91 的官方重放结果确认这段实际锤子运动未跟准参考：仅 21/99 步达标，末步仍偏离，最大位移误差约 0.084 m，超过 0.05 m 上限。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频 0.40、1.00、2.00、3.96 s 的两个视角显示，MyoHand 的接触集中在锤柄末端，锤头在较长力臂下悬垂，缺少能明确约束锤子朝向的对向夹持证据；这种勾持几何不利于控制整把锤子的位姿。agent 在 log step 95 也自述最佳方案只勾住柄端、锤头拖动或悬挂；这是与视频一致的自述，支持将跟踪失败归于接触设计，但没有验证过可成功替代的握法。"
+   }
+  ],
+  "gap_zh": "观察：inputs/ 没有状态日志或成功 oracle；视频不能精确恢复逐步质心误差、相对旋转误差及接触力，故不能确定首次误差越界的时刻。最终文件的官方重放摘要见 log step 91；native_metrics 的 track_ok=21 和 track_max≈0.084 m 取自其 reason，后者仅有三位小数；log step 90 的末步 info 被截断，reward 也未提供末步位置和旋转误差，故这两项 value 留 null。根因中的承载效果是从可见勾持几何与 agent 自述综合推断，具体手指接触对未记录。"
+ },
+ "summary_zh": "MyoHand 以弯曲手指勾住锤柄末端并带起锤子，但未完成「沿参考挥锤」，最终节点为 0/1。视频 1.20–3.96 s 显示锤头悬垂；log step 91 对最终文件的两次官方重放均确认仅 21/99 步达标、末步未达标、最大位移误差约 0.084 m，接触设计不足是有证据支持的根因推断。没有状态日志或成功参考视频，无法精确定位首次超限时刻，也无法核实末步两项误差及接触力。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "单节点未达成、最终 0/1、breakpoint 与 log step 90/91 的官方重放结果（21/99 步达标、末步偏离、最大误差约 0.084 m）一致；视频所见勾持锤柄末端、锤头悬垂，位姿不准与接触设计根因均有证据且标为推断，时钟未换算。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:00:06Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--ugrcnag/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-privileged--ugrcnag/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "bf1cbdb2dbc116d93d94aedd1beed6b2f4d2b372f4bffd2c9e0c41210e4182ff",
+   "video_sha256": "a86825ddd05d4fb68c44ccc6e44b4a7c78ef09baa943ca7a736b40191874088c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--5y2nmqj.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--5y2nmqj.json
new file mode 100644
index 0000000000000000000000000000000000000000..ecb2ed67f49c198333953d202b3c1a0efe82581b
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--5y2nmqj.json
@@ -0,0 +1,224 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-standard--5y2nmqj",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 53,
+   "sim_steps": 53,
+   "physics_steps": 530,
+   "steps": 550,
+   "calls": 32,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 550,
+   "elapsed_s": 2425.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.023,
+   "peak_processes": 13,
+   "peak_disk_gb": 1.412,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-standard--5y2nmqj/replay.mp4",
+  "duration_s": 2.16,
+  "fps": 25.0,
+  "frames": 54,
+  "sha256": "4564fa07b3bc6d3014eb6367a85462ecec385f3c5de35213116ea67ebcbc4b64"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.56,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.56 s 的两个视角中，MyoHand 的手指从桌面锤柄上方抬开，棕色锤柄与白色锤头仍留在桌面，未随手抬起。"
+     },
+     {
+      "t": 1.2,
+      "log_step": null,
+      "text_zh": "观察：视频 0.72–1.40 s，MyoHand 屈指靠近锤柄端部，指端贴到锤柄上侧；锤柄未被手包住并带离桌面。视频 1.44–2.12 s 的连续帧中，手与锤柄进一步分离，末帧锤子仍在桌面。"
+     },
+     {
+      "t": null,
+      "log_step": 203,
+      "text_zh": "观察：log step 203 的机器人服务返回先重置到 episode=11、resets_left=0，再执行最后一局；该局 control_steps=53，done=True，reason 为锤子距参考或腕部超过 0.25 米。inputs/run.json 的评分器字段 graded_by_live=1、n_actions=53、success=0，确认该局未完成要求的 99 个控制步。"
+     },
+     {
+      "t": null,
+      "log_step": 203,
+      "text_zh": "推断：结合视频中锤子始终未随手抬起、log step 203 确认评分局在第 53/99 个控制步提前结束，整段参考挥锤验收从未成立；初始位姿接近参考不计为完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.08 s 及视频约 0.96–1.36 s，MyoHand 指端贴压棕色锤柄的上侧或端部；屈指后没有把锤柄包夹在手中，随后手离开而锤子仍留桌面。实际接触记为指端压擦锤柄。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 1.2,
+   "log_step": 203,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 1.00–1.40 s，MyoHand 已闭合手指并运动，但锤子仍在桌面，未被带起；到视频 2.12 s 也没有修复。log step 203 的服务结果确认最后评分局在第 53/99 个控制步因锤子距参考或腕部超限而结束。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 1.2,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.72–1.40 s，MyoHand 的手指在锤柄端部上方屈曲并与锤柄贴碰，但没有形成能带起锤子的包夹；视频 1.44–2.12 s，手离开锤柄而锤子仍贴桌面。所缺的是持锤接触，未见先握持再掉落的阶段。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "t": 1.2,
+    "log_step": 201,
+    "evidence_zh": "推断：log step 201 中实际用于最后一局的脚本，以固定的张指下压、屈指、抬臂序列尝试夹锤；视频 0.96–1.36 s 显示手指在锤柄端部上侧闭合，未形成包住锤柄的对向接触，后续抬臂没有带起锤子。这支持抓取接触布局未能承载锤子；没有证据验证另一种握法能否成功。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "没有本局状态日志、接触力或逐步轨迹误差记录，故不能精确判断各指接触力，也不能还原 track_pos_err、track_rot_err、track_ok、track_max，四项原生指标均留 null；服务的合并终止原因不能区分锤子距参考超限还是距腕部超限。无成功 oracle，替代握法的可行性未知。视频时长为 2.16 s，而服务第 53 个控制步按 50 Hz 对应仿真 1.06 s，未确认两者时钟映射；所标视频时间均直接取自已检查的帧，没有把日志控制步当成视频秒数。",
+  "unmet": {
+   "hammer_reference_followed": false
+  }
+ },
+ "summary_zh": "最后评分的第 11 局中，MyoHand 尝试屈指夹持并抬臂，但锤子始终留在桌面；唯一节点“沿参考挥锤”未达成，终态为 0/1。失败发生在持锤接触未建立，视频支持抓取接触布局不适配，服务确认该局在第 53/99 个控制步提前终止；缺少状态和原始误差记录，无法量化轨迹误差或确定具体超限距离。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点未达成、锤子始终留在桌面、断点与现象/根因均有帧和服务读数支持；未发现需要改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:08:48Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--5y2nmqj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--5y2nmqj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f6115bff8d712ae28379439fd719233ea176a572144f8bb7ac2ff09fee71755a",
+   "video_sha256": "4564fa07b3bc6d3014eb6367a85462ecec385f3c5de35213116ea67ebcbc4b64",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--bd7dafh.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--bd7dafh.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec77bc2849153282b626d89df000293fa1df5b96
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--bd7dafh.json
@@ -0,0 +1,217 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-standard--bd7dafh",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 25,
+   "sim_steps": 25,
+   "physics_steps": 250,
+   "steps": 264,
+   "calls": 295,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 264,
+   "elapsed_s": 3586.6,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.087,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.623,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-standard--bd7dafh/replay.mp4",
+  "duration_s": 1.04,
+  "fps": 25.0,
+  "frames": 26,
+  "sha256": "a93a7433257962b1070950611d481389368b33d479607b77d0671c6dc2819708"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：逐帧查看视频 0.00–1.00 s 的全部 26 帧及双视角局部；MyoHand 手指在锤柄末端上方屈曲，视频 0.64 s 时锤子仍平放在桌面，未见手将锤子提起。"
+     },
+     {
+      "t": 0.88,
+      "log_step": 95,
+      "text_zh": "观察：视频 0.80–1.00 s，MyoHand 闭合的手与锤柄逐渐分离，锤子仍在桌面，未随手臂挥动；log step 95 的机器人服务最终状态为 episode=11、control_steps=25、done=True，原因是物体距参考或腕部超过 0.25 m。"
+     },
+     {
+      "t": 1.0,
+      "log_step": 95,
+      "text_zh": "推断：log step 92 的最后一次 reset 后，log step 93–95 依次执行抓取与后续轨迹脚本，最终服务状态的 25 个控制步与评分器 n_actions=25 相符；因此本视频对应第 11 个 episode，而非此前重置掉的测试。该 episode 在完成要求的 99 个控制步前已被原生终止，整段沿参考挥锤节点从未成立。视频锚点直接取自帧时间，未将 agent 时间或控制步换算成视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "推断：视频 0.32–0.68 s 的两视角显示 MyoHand 手指贴着锤柄末端上侧屈曲、擦过，随后在视频 0.80–1.00 s 向上离开；未形成可带起锤子的对向夹持，故实际接触按压/擦记录。锤子自身为受评对象。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 0.8,
+   "log_step": 95,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.64–0.80 s，MyoHand 手指在锤柄上方闭合，手开始上移而锤子没有随之离桌；视频 0.88–1.00 s 手锤间隔继续增大，未恢复抓持。log step 95 的机器人服务报告最后的第 11 个 episode 在第 25 个控制步因物体距参考或腕部超过 0.25 m 提前结束，未完成沿参考挥锤。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.32–0.68 s，MyoHand 手指在锤柄末端上侧屈曲；视频 0.80–1.00 s，闭合的手向上离开，锤子仍平放桌面。所需的抓持接触未建立；未见先持锤离桌再滑落的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频 0.48、0.64、0.72 s 的俯视与侧视显示 MyoHand 闭合手指的位置偏在锤柄末端上方，未形成围住锤柄的对向承载接触；视频 0.80–1.00 s 手移开而锤子留在桌面，说明这次手指闭合与锤柄的接触布局没有建立可用于抬锤的抓持。没有成功对照或接触力记录，不能验证可行替代握法。"
+   }
+  ],
+  "gap_zh": "无状态日志或成功 oracle，视频分辨率及肌腱线条遮挡使逐指触点、接触力和瞬时轻触只能结合前后帧推断；未见有效抓持。评分器未提供 track_pos_err、track_rot_err、track_ok、track_max 原始值，不能量化轨迹误差或合格步数；服务终止原因亦未区分距参考超限与距腕部超限。所有视频时间直接来自所查看帧，未假设其与仿真时间等速。"
+ },
+ "summary_zh": "最后受评的第 11 个 episode 中，MyoHand 在锤柄末端上方屈指并移开，锤子始终未被有效带起，沿参考挥锤节点未达成（终态 0/1）。断点在视频约 0.80 s 的抓持操作中，接触布局未形成有效夹持；log step 95 的机器人服务确认第 25 个控制步距离超限而提前结束。缺少状态日志、成功对照及原始轨迹误差，不能进一步核实逐指受力和精确跟踪指标。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频 0–1.04 s 与控制步一一对应（25 步），锤子始终留在桌面、手指在柄端上方屈曲后移开、参考球上移，节点未达成、断点与「没建立接触」/「抓取和接触设计」均有帧证据支持，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:07:20Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--bd7dafh/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--bd7dafh/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b6677d14659af9d0a4f2d76651e64b383f58d429a8c64619c6b71130d770765d",
+   "video_sha256": "a93a7433257962b1070950611d481389368b33d479607b77d0671c6dc2819708",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--f6onwkp.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--f6onwkp.json
new file mode 100644
index 0000000000000000000000000000000000000000..c6c439e58fd8650db2b8ade669f6158f62cb2908
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--f6onwkp.json
@@ -0,0 +1,217 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-standard--f6onwkp",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 99,
+   "sim_steps": 99,
+   "physics_steps": 990,
+   "steps": 243,
+   "calls": 6,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 243,
+   "elapsed_s": 2694.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.557,
+   "peak_processes": 16,
+   "peak_disk_gb": 0.411,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-standard--f6onwkp/replay.mp4",
+  "duration_s": 4.0,
+  "fps": 25.0,
+  "frames": 100,
+  "sha256": "ad47e4798ec97152b3b3a27b532a1852843718182743fda14bf916d2aceef0dc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.6,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.60 s，MyoHand 的手指在锤柄末端弯曲接触，锤子随手部动作离开原先平放姿态；视频 0.80–1.40 s 锤柄相对手掌明显翻转，未形成持续随手挥动的稳定关系。"
+     },
+     {
+      "t": 2.12,
+      "log_step": null,
+      "text_zh": "观察：视频 2.08–2.40 s，锤柄与 MyoHand 手指逐渐分离；视频 3.00–3.96 s，手仍在动作，锤柄位于手的下方或侧下方，未重新进入手指围合范围。"
+     },
+     {
+      "t": 3.96,
+      "log_step": 67,
+      "text_zh": "观察：log step 67 是最后一次提交：agent 的脚本先重置已结束的上一局，再提交 99 行执行器命令；该次服务终局 reason 报告“on track at 20 of 99 steps, off track at the end, 0.240 m off at worst”。这与视频末段的手锤分离一致，判定整段沿参考挥锤节点从未完成；此前 log step 61 的 6/99 是已被重置的另一局。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.00–0.60 s，单只 MyoHand 的拇指与弯曲手指围合锤柄末端，短暂带动锤子；随后锤柄相对掌部翻转、滑离，视频 2.12 s 后手锤分开。此处标短暂抓持，不表示整段稳定持锤。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": 20
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": 0.24
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 2.12,
+   "log_step": 67,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.80–1.40 s 锤柄已相对 MyoHand 掌部大幅翻转；视频约 2.12 s 锤柄脱离手指，至视频 3.96 s 未重新抓回。log step 67 的最终评分局完成 99 个控制步，但服务报告仅 20 步达标、末步偏离、最大位置误差 0.240 m；失败发生在沿参考挥锤的操作中。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.12,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.60 s 锤柄尚在 MyoHand 弯曲手指与拇指附近并随手移动；视频 0.80–1.40 s 锤子在掌旁翻转，视频 2.08–2.40 s 锤柄滑出手指范围，后续至视频 3.96 s 手部继续动作而锤子留在手外，没有恢复抓持。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频 0.60–1.40 s，MyoHand 的手指保持弯曲，但锤柄没有固定在指间，而是相对掌部翻转到腕旁；视频 2.08–2.40 s 后完全脱离。现有围合位置未能持续约束锤柄的转动和滑移，导致随后的参考运动失去传力接触。agent 在 log step 68 亦自述始终没有获得牢固握持，但该自述仅作支持；现有证据未验证可行的替代抓法。"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功 oracle；视频可见短暂抓持及后续手锤分离，但手指接触区有重叠，缺少最终评分局逐步接触对、接触力及未舍入的位置/旋转误差，无法确定滑移最初发生的精确时刻或具体肌肉参数原因。track_ok=20 和 track_max=0.240 m 取自 log step 67 的最终服务判词，后者为显示精度；末步 track_pos_err、track_rot_err 未报告，保留 null。视频长 4.00 s，而任务的 99 控制步为仿真 1.98 s，未核实逐帧对应关系，所有导航时间均直接取视频帧，未按控制步换算。"
+ },
+ "summary_zh": "评分尝试短暂带动锤子抬起并翻转，但未保持抓持，视频约 2.12 s 后手锤分离且未恢复，沿参考挥锤节点最终为 0/1。log step 67 的最终服务判词给出 20/99 步达标、末步偏离、最大位置误差 0.240 m，支持节点未完成；失败归于持锤接触未能持续约束滑移与转动。缺少状态日志和逐步误差/接触力，无法核实更细的控制原因及滑移起点。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点 ref 与 2.12 s 视频锚点、接触/没保持住与抓取和接触设计根因均与帧和 log step 67 的服务判词（20/99、0.240 m）一致，未发现需改内容的错误；视频为交错双视角，手锤分离的判断可由帧支持。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:09:49Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--f6onwkp/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--f6onwkp/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "855cca13e3db307a1c4ac448152b4baf97584abf9c2b4394f343eb6b8fe71998",
+   "video_sha256": "ad47e4798ec97152b3b3a27b532a1852843718182743fda14bf916d2aceef0dc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--i7jq3rs.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--i7jq3rs.json
new file mode 100644
index 0000000000000000000000000000000000000000..f90e1493edf45e0cf7d0547f6648e56b46c9d0ae
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--i7jq3rs.json
@@ -0,0 +1,217 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-standard--i7jq3rs",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 44,
+   "sim_steps": 44,
+   "physics_steps": 440,
+   "steps": 223,
+   "calls": 100,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 223,
+   "elapsed_s": 2019.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.442,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.254,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-standard--i7jq3rs/replay.mp4",
+  "duration_s": 1.8,
+  "fps": 25.0,
+  "frames": 45,
+  "sha256": "13a1c34eca01999467f47fc83305df039a53a63b4ccfe1db2b3f18e47cd0df28"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.20 s 的侧视中，MyoHand 指端贴近木制锤柄上表面后向内卷曲；视频 0.24–0.64 s 闭合的手指位于锤柄上方，二者之间出现清晰间隙，没有看到锤柄被拇指与其余手指夹在中间。"
+     },
+     {
+      "t": 1.76,
+      "log_step": null,
+      "text_zh": "观察：已逐帧检查视频 0.00–1.76 s 的全部 45 帧，包含最后一秒；后段锤子与手继续分离，视频 1.76 s 锤子在手的左上方，手仍空握，未见手持锤完成往复挥动。"
+     },
+     {
+      "t": null,
+      "log_step": 152,
+      "text_zh": "观察：log step 147 显示最终脚本重置至 episode 11 后提交 99 行动作；log step 152 的该次结果为 control_steps=44、sim_time_s=0.88，log step 153 为 episode=11、executed_rows=44，与评分器 n_actions=44、live_success=0 一致。最终评分局没有执行满所需的 99 个控制步；仿真时长不直接作为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频 0.00–0.16 s，MyoHand 指端从上方贴触锤柄表面，随后卷曲离开；没有可见的对向夹持，不计为抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 0.24,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.24 s，MyoHand 手指已在锤柄上方卷曲成空握，锤柄没有进入指间；视频 0.64–1.76 s 未恢复为持锤状态。log step 152–153 所记录的最终 episode 11 仅执行 44/99 个控制步，沿参考挥锤未完成。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.24,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–0.16 s 手指仅贴触锤柄上方；视频 0.24–0.64 s 手指闭合后与锤柄有间隙，后续至视频 1.76 s 仍未见锤柄被夹持。推断：失败的是所需的持锤接触没有建立，未见先稳定握住再滑脱的过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 144,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "观察：log step 144 的最终脚本给手指屈肌及拇指肌持续激活，并预先生成抓取阶段与参考挥动阶段的 99 行指令，以单次 submit 执行，阶段切换没有根据是否夹住锤柄作调整；log step 147 确认该脚本提交到最终 episode 11。视频 0.24–0.64 s 却显示手指在锤柄上方空握，至视频 1.76 s 也未建立持锤接触。推断：这套控制未把夹持指令实现为有效持锤，且空握后继续执行预定动作；现有证据不足以确定更细的控制子类。"
+   }
+  ],
+  "gap_zh": "未提供状态日志及成功 oracle；视频可判断手指与锤柄分离，但没有逐步接触对、接触力或物体位姿记录，不能区分具体接触几何、肌肉响应和轨迹执行问题。评分器未报告 track_pos_err、track_rot_err、track_ok、track_max，四项指标均填 null。log step 152–153 只展示最终结果的部分字段，没有 episode 11 的具体终止原因；log step 137、147 中的距离终止原因属于 episode 10，不移用为最终局原因。最终局仿真 0.88 s 与视频 1.8 s 不同，所有视频锚点均直接按帧取时，未将两者换算。"
+ },
+ "summary_zh": "评分局为最终 episode 11：MyoHand 指端短暂贴触锤柄后在其上方闭合，未建立持锤接触，沿参考挥锤节点未达成，最终为 0/1。断点约在视频 0.24 s 的空握，控制指令未实现有效夹持；该局仅执行 44/99 个控制步。没有状态日志及逐步误差指标，无法确定更细控制原因或最终原生终止原因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点在持锤接触未建立、contact_actual（压/擦，非抓握）与视频一致；时钟未换算，final=0 与评分器一致，无可确认的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:10:11Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--i7jq3rs/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--i7jq3rs/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "04fb386ea3435de34205d2d10a52fda99501d6f03e85e18743928d80de4cc55a",
+   "video_sha256": "13a1c34eca01999467f47fc83305df039a53a63b4ccfe1db2b3f18e47cd0df28",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--iguudjt.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--iguudjt.json
new file mode 100644
index 0000000000000000000000000000000000000000..c2cd8d53ce68f3a2b549fb31ab7c780eeeb0ea91
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--iguudjt.json
@@ -0,0 +1,241 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-standard--iguudjt",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 51,
+   "sim_steps": 51,
+   "physics_steps": 510,
+   "steps": 51,
+   "calls": 3,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 51,
+   "elapsed_s": 3358.4,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.599,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.371,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-standard--iguudjt/replay.mp4",
+  "duration_s": 2.08,
+  "fps": 25.0,
+  "frames": 52,
+  "sha256": "eaa81e36fef93339f191532c25d7b389397ba447bdeeedf13eb6e5f5f599d133"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.64,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.80 s 的两个视角显示，MyoHand 手指弯曲靠近锤柄上侧，锤子仍由桌面承托；视频 0.56、0.64、0.72 s 的近看帧未见拇指与其他手指从两侧夹住锤柄。"
+     },
+     {
+      "t": 0.96,
+      "log_step": null,
+      "text_zh": "观察：视频 0.88–2.04 s，MyoHand 手腕和弯曲手指逐渐离开锤柄，手与锤柄之间出现明显间隙；锤子一直平躺在桌上，没有被带起挥动。已检查全部 52 帧，包括视频最后一秒。"
+     },
+     {
+      "t": null,
+      "log_step": 86,
+      "text_zh": "观察：log step 84 的部署脚本生成 99 行动作并在 reset 后提交；log step 86 的实际服务 status 返回 episode=2、control_steps=51、done=True，reason 为锤子距参考或手腕超过 0.25 m 而失败。inputs/run.json 的评分器字段也记录 steps=51、live_success=0、task_failure=1，未执行完要求的 99 个控制步。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：视频约 0.56–0.80 s，MyoHand 弯曲手指的指尖在桌面锤柄上侧短暂压触，未见包围锤柄的对向夹持；视频约 0.88 s 后手指与锤柄分离，空手抬升。锤子自身为受评对象。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 0.88,
+   "log_step": 86,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.88 s，MyoHand 开始离开锤柄而锤子仍平躺在桌面，之后直至视频 2.04 s 未补成抓握或带起锤子；最终提交回合的服务在 log step 86 报告第 51 个控制步因锤子距参考或手腕超过 0.25 m 提前失败。断点取未能夹持并带动锤子的挥锤步骤。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 0.88,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.56–0.80 s，机器人弯曲指尖仅在锤柄上侧接触，没有形成夹持；视频 0.88–2.04 s，手指离开锤柄且锤子一直留在桌面。所需的承载抓握从未建立，没有可见的持锤后滑脱过程。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "alt_feasible": "未知",
+    "t": 0.88,
+    "log_step": 86,
+    "source_zh": "视频 / 日志",
+    "evidence_zh": "推断：视频 0.56–0.80 s，MyoHand 弯曲手指靠近桌面上的锤柄，未见形成承载夹持；视频 0.88–2.04 s，手腕和手指离开锤柄，锤子没有随手运动而仍留在桌面。log step 84 的部署脚本将本地生成的动作序列开环提交，log step 86 的最终评分服务返回 episode=2、control_steps=51、done=True，reason 为物体距参考或手腕超过 0.25 m。结合手运动而锤子未被带动的现象，将根因归为控制；抓取和接触设计不适合、轨迹或逆解偏差、反馈控制不足均为竞争解释，不能仅凭手腕离开就确认轨迹跟随失稳。抓握设计及替代握法是否可行未验证，服务 reason 也未区分参考距离与腕部距离哪一项超限。"
+   }
+  ],
+  "gap_zh": "评分回合没有状态日志、逐步接触力或原始轨迹误差记录，也没有成功 oracle；可从视频确认未形成承载抓握，但不能确定精确的手腕朝向偏差或闭合参数。评分器 reward 未报告 track_pos_err、track_rot_err、track_ok、track_max，故四项 native_metrics 均为 null；log step 79–82、86 中本地 replica 的误差数值不作为评分回合的原生指标。视频为 52 帧、25 fps、时长 2.08 s，而服务报告的 51 个控制步按任务控制周期 0.02 s 为仿真 1.02 s，未建立精确映射；所有视频锚点直接取已查看帧，不把控制步或 agent 工作时间换算为视频时间。 根因子类未能区分：视频不能确定手腕是在按参考抬升还是发生跟随偏差，抓取和接触设计、轨迹或逆解、反馈控制仍是竞争解释。log step 82 的本地 replica 接触诊断可作会话级根因辅助证据（docs/03_trajectory.md:225–226），但不是评分回合的逐步接触读数，也未验证抓握方案可行性，故不足以确定控制子类。log step 86 的服务终止原因是参考距离或腕部距离超限的析取，不能据此单独认定腕部漂移。"
+ },
+ "summary_zh": "MyoHand 已尝试接触锤柄并移动手腕，但锤子始终留在桌面，唯一节点“沿参考挥锤”全程未达成，最终为 0/1；视频约 0.88 s 起手与锤柄分离，最终评分服务在 log step 86 报告第 51 个控制步因物体距参考或手腕超过 0.25 m 提前失败。根因收敛为“控制·控制”：抓取和接触设计、轨迹或逆解、反馈控制尚不能区分，抓握设计及替代方案可行性未验证，因这一关键归因缺口将 conf 从 8 降至 5。评分回合缺少逐步状态与原始误差，四项原生指标仍无法填值。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "WHERE 标签（节点未达成、断点 hammer_reference_followed、时钟均未换算）与视频和服务读数一致；问题在根因：视频可见的是手腕/手指整体逐渐漂离锤子，标注却把根因定为“抓取和接触设计”，该推断缺乏支持，更像轨迹跟随/反馈控制问题。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、时钟（未换算控制步）与视频及服务读数一致；根因已按上轮收敛为“控制·控制”并如实说明竞争解释，无新增错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:13:40Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--iguudjt/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--iguudjt/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--iguudjt/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--iguudjt/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0cffac1059e67565e6e557cafc35a31967f510511127975edbaf73bb41903cbf",
+   "video_sha256": "eaa81e36fef93339f191532c25d7b389397ba447bdeeedf13eb6e5f5f599d133",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--jebaa5f.json b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--jebaa5f.json
new file mode 100644
index 0000000000000000000000000000000000000000..06c7bdc6b2e85ac83574167cd59cd241e939e723
--- /dev/null
+++ b/phase2/myosuite/dex_hammer_use/trajectories/myosuite-dex-hammer-use-i00-standard--jebaa5f.json
@@ -0,0 +1,238 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-dex-hammer-use-i00-standard--jebaa5f",
+ "taskKey": "myosuite/dex_hammer_use",
+ "dag_sha256": "925558ce4c1e26b62ae98f9f8bb9f72df3dd2873a50bb648f9d648a4041f166c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 50,
+   "sim_steps": 50,
+   "physics_steps": 500,
+   "steps": 263,
+   "calls": 271,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 263,
+   "elapsed_s": 2096.9,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.585,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.851,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-dex-hammer-use-i00-standard--jebaa5f/replay.mp4",
+  "duration_s": 2.04,
+  "fps": 25.0,
+  "frames": 51,
+  "sha256": "b1dae4cbf5032a947ecb1363858a904784a4be723e3529eded9ac900f7fff0dc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hammer_reference_followed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.24,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–0.40 s，MyoHand 单手在锤柄上方屈指；视频 0.24 s 可见拇指侧指端抵住柄的上表面，其余手指没有从另一侧包住锤柄。"
+     },
+     {
+      "t": 1.68,
+      "log_step": null,
+      "text_zh": "观察：视频 0.48–1.00 s，MyoHand 屈曲的手指在锤柄上方移动，锤子仍接近桌面；视频约 1.24–1.60 s，锤柄部分被手遮挡；视频 1.68、1.84、2.00 s，锤子在手下方及旁侧抬起并倾斜，随手移动，但与白色半透明参考锤的位置和朝向明显错开。已检查全片及最后一秒的逐帧画面，不能再据此声称手锤持续分离或锤子始终贴桌；是否形成对向抓握无法确认。"
+     },
+     {
+      "t": 2.0,
+      "log_step": 121,
+      "text_zh": "观察：log step 93 重置进入第 11 局，log step 121 的服务状态仍为 episode=11，control_steps=50、done=True，reason 为锤子距参考或手腕超过 0.25 米；这局提前终止，未执行节点要求的全部 99 个控制步。视频末帧为视频 2.00 s；此处分别引用视频时钟与服务控制步数，不把 agent 工作时间换算为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "未知",
+      "note_zh": "观察：视频 0.24–0.32 s，MyoHand 的拇指侧指端抵住锤柄上表面，呈压/擦接触；视频 1.68–2.00 s，锤柄位于屈曲手指下方及旁侧，锤子随手移动而抬起、倾斜。推断：后段存在手与锤子的接触带动，但接触区域被手和锤柄遮挡，无法确认对向夹持或区分抓握、压/擦等具体方式，因此整条接触链的 how 记为未知。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "track_pos_err",
+   "unit": "m",
+   "threshold": 0.03,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_rot_err",
+   "unit": "rad",
+   "threshold": 0.5,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "track_ok",
+   "unit": "control_steps",
+   "threshold": 90,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "track_max",
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hammer_reference_followed",
+   "t": 1.68,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 1.68–2.00 s，MyoHand 已带动锤子抬起、倾斜，但锤子与白色半透明参考锤的位置和朝向仍明显错开，未恢复为沿参考运动。log step 93 重置后为第 11 局；log step 121 的同局服务状态记录 control_steps=50、done=True，因锤子距参考或手腕超过 0.25 米而提前结束，未完成节点要求的全部 99 个控制步。视频时刻与服务控制步数分别引用，不换算 agent 工作时间。"
+  },
+  "unmet": {
+   "hammer_reference_followed": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.68,
+    "log_step": null,
+    "evidence_zh": "观察：视频 1.68、1.84、2.00 s，MyoHand 附近的锤子已经抬起并改变倾角，但实物锤柄及锤头与白色半透明参考锤明显错位，运动未跟随参考；这不是锤子始终未受手带动的情形。log step 121 的第 11 局服务状态报告因锤子距参考或手腕超过 0.25 米而终止。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 93、99 记录第 11 局执行接近与闭指控制，log step 106、109、111 改用 grip=-1，log step 117、121 改回 grip=0.45；视频 1.68–2.00 s 显示手已带动锤子抬起、倾斜，却未使其沿白色半透明参考锤运动。结合 log step 121 的提前终止服务状态，控制未能协调手指与腕部运动以实现参考跟踪；现有证据无法确定问题属于抓取和接触设计、轨迹或逆解、反馈控制还是参数调试，根因保留在控制。",
+    "source_zh": "视频 / 日志"
+   }
+  ],
+  "gap_zh": "没有状态日志或成功参考视频，现有视频和服务输出未报告 track_pos_err、track_rot_err、track_ok、track_max，因此四项原生指标均为 null。视频后段可见锤子随手移动并倾斜，但手与锤柄遮挡了具体接触区域，无法确认是否形成对向抓握、是否曾滑移及具体接触链接或接触力；不能据此认定没建立接触或没保持住。缺少能区分抓取和接触设计、轨迹或逆解、反馈控制与肌肉参数问题的独立证据，根因保留在控制。"
+ },
+ "summary_zh": "受评的是最后第 11 局，MyoHand 曾以指端抵住锤柄，视频后段也带动锤子抬起并倾斜，但锤子的运动与参考明显错位，唯一节点未达成（0/1）。断点为沿参考挥锤的操作中，现象是位姿不准；服务在第 50 个控制步因锤子距参考或手腕超过 0.25 米而提前终止。具体抓握方式与控制子类因遮挡和原始状态缺失仍不确定，置信度调整为 7。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点 ref 与时钟都没问题；但标签称视频后半段手与锤子分离、锤子一直贴桌，与帧不符：约 1.6–2.0 s 锤子被手带得抬起并倾斜。「没建立接触」的现象因此缺乏支持，需改写。",
+    "n_issues": 3,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "第2轮修订后标签与画面一致：唯一节点未达成，锤子后段被手带动但与参考错位，位姿不准、断点和控制根因（子类未知）均有证据支持；时钟分别引用视频与控制步数，无换算。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:14:06Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--jebaa5f/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--jebaa5f/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--jebaa5f/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-dex-hammer-use-i00-standard--jebaa5f/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "dce5f317bb54346076ac2cef31b58ff7bf94ef9439c76b39f6a39f093b4a7cb4",
+   "video_sha256": "b1dae4cbf5032a947ecb1363858a904784a4be723e3529eded9ac900f7fff0dc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--jg72q27.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--jg72q27.json
new file mode 100644
index 0000000000000000000000000000000000000000..703c6f665b3b75255022d3ac12370616427be080
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--jg72q27.json
@@ -0,0 +1,188 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-privileged--jg72q27",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 200,
+   "sim_steps": 200,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 30,
+   "agent_resets": 36047,
+   "agent_sim_steps": 6653912,
+   "agent_records": 0,
+   "agent_replays": 3,
+   "agent_cost_self_reported": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-privileged--jg72q27/replay.mp4",
+  "duration_s": 8.04,
+  "fps": 25.0,
+  "frames": 201,
+  "sha256": "f0b46c8b6d5a96fa67051dc1e42085fac43bfd452c234e94ac0fab65daebb86e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 实骰已在 MyoHand 掌指间，但其彩色面和棱边朝向与旁侧半透明目标骰子不同，目标周围成功指示为红色，初态不满足重定向节点。"
+     },
+     {
+      "t": 2.4,
+      "log_step": null,
+      "text_zh": "观察：视频 1.20–4.80 s 的双视图中，MyoHand 手指收拢并拨转实骰，实骰随之改变朝向和在手中的位置；实际执行了掌内调整，红色成功指示没有转绿。"
+     },
+     {
+      "t": 8.0,
+      "log_step": null,
+      "text_zh": "观察：已检查视频全程每 0.08 s 的目标指示局部帧，以及视频 7.04–8.00 s 的密集双视图；指示始终红色，末帧实骰仍在掌指旁，彩色面棱边仍与目标有偏角。成功参考视频 6.00–9.00 s 的同一指示明显为绿色，仅用来识别达标外观；本次未见该线索。"
+     },
+     {
+      "t": null,
+      "log_step": 201,
+      "text_zh": "观察：log step 200 保存的是 200 控制步轨迹，服务终态 solved=false、solved_run=0、done=false；log step 201 对该交付文件两次新进程重放均 success=false，终态哈希相同，且与评分器 deterministic=1、sim_steps=200、task_failure=0 一致。log step 202 的追加控制测试在控制步 239 掉落，但未覆盖保存文件，不属于本次评分视频。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 1.20–4.80 s，单只 MyoHand 的拇指与其余手指围住掌中实骰，收拢、拨转并改变其朝向；视频末帧实骰仍处于掌指间，未见计划松手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": 4.8,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 4.80 s，MyoHand 完成主要拨转后，实骰停在仍有朝向偏差的掌指间位姿；之后至视频 8.00 s 末帧未修正达标，成功指示持续红色。该步已实际操作，未达目标位姿及连续保持验收。"
+  },
+  "unmet": {
+   "die_at_target_pose": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 4.8,
+    "log_step": 196,
+    "evidence_zh": "观察：视频 4.80–8.00 s，手指保持实骰，但实骰的彩色面与棱边仍相对目标倾斜，目标指示一直红色。log step 196 中候选的终态位置误差约 0.03742 m、欧拉角误差范数约 0.36939 rad，均超过阈值；该候选终态哈希 f01f324117d1792f 与 log step 200 保存、log step 201 两次交付文件重放一致，支持其残余位姿偏差。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：agent 的肌肉控制序列未能在拨转骰子的同时将其位置与朝向同时收敛到允许范围。视频 1.20–4.80 s 显示确有手指拨转，视频 4.80–8.00 s 仍留偏差；log step 196 的匹配终态误差、log step 200 保存时 solved_run=0 及 log step 201 确定性重放失败支持这一控制归因，无法进一步区分接触布局与具体肌肉控制问题。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "观察：inputs/ 中没有模拟状态日志，评分器 reward 未报告 pos_dist、rot_dist、solved_run，故三个 native_metrics 的 value 均留 null；log step 196 和 log step 200 的匹配交付终态读数只作为补充证据。无法重建逐控制步误差、接触对和接触力，不能细分控制根因或给出精确数值越界时刻；断点视频 4.80 s 是主要拨转后仍留偏差的近似导航锚点。"
+ },
+ "summary_zh": "观察：MyoHand 实际在掌指间拨转了骰子，但视频全程成功指示均为红色，末帧仍有位姿偏差，“骰子转到目标”始终未达成，最终为 0/1。推断：失败发生在该步操作中，现象为位姿不准，匹配交付文件的终态误差与确定性重放结果支持控制归因；追加测试中的掉落未包含在交付轨迹。观察：没有逐步状态与接触力日志，无法进一步细分控制问题。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频、日志一致：实骰在手中被拨转但目标指示全程红色，节点未达成，final=0；断点、位姿不准现象及控制根因均有视频与 log step 196/200/201 支持，追加测试掉落未计入。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:51:50Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--jg72q27/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--jg72q27/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "89274bfa4cd3b922965f755faa2bd2babad4124866f739bd68f87a9a11275ead",
+   "video_sha256": "f0b46c8b6d5a96fa67051dc1e42085fac43bfd452c234e94ac0fab65daebb86e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--khafnxf.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--khafnxf.json
new file mode 100644
index 0000000000000000000000000000000000000000..fa26c76c1765cf46a34e201f4321cdc88238d011
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--khafnxf.json
@@ -0,0 +1,176 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-privileged--khafnxf",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 246,
+   "sim_steps": 246,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 110,
+   "agent_resets": 18736,
+   "agent_sim_steps": 2268511,
+   "agent_records": 0,
+   "agent_replays": 76,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.003,
+   "peak_processes": 19,
+   "peak_disk_gb": 1.007,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-privileged--khafnxf/replay.mp4",
+  "duration_s": 9.88,
+  "fps": 25.0,
+  "frames": 247,
+  "sha256": "0465a996ada8589134f51b8bc2d8ff762c45f39b5e78b20a112b1ec054eaea93"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [
+     [
+      9.84,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s，实骰已在 MyoHand 掌指间，彩色面朝向与旁侧半透明目标不同，目标周围成功指示呈红色，整步初始未达成。"
+     },
+     {
+      "t": 0.8,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–2.40 s，MyoHand 手指屈曲并围住实骰，骰子随掌指接触转动，红色面与绿色面的朝向改变；视频 5.40–5.84 s 骰子仍夹在掌指间，指示仍红。"
+     },
+     {
+      "t": 5.88,
+      "log_step": null,
+      "text_zh": "观察：视频 5.84 s 指示仍红，视频 5.88 s 变绿；随后直至视频 9.84 s，实骰留在屈曲手指与掌部之间，指示保持绿色，未见脱手下落。"
+     },
+     {
+      "t": 9.84,
+      "log_step": 179,
+      "text_zh": "推断：log step 179 对交付动作文件的等值再生轨迹调用基准服务 replay，返回 steps=246、solved_run=100、success=True，且动作与交付文件最大差为 0；log step 239 官方双重重放同为 246 步成功且终态哈希相同。视频共 247 帧、25 fps，对应初帧及 246 个控制步；已用视频 5.88 s 变绿事件核对第 147 控制步，连续合格第 147–246 步共 100 步，于视频 246/25=9.84 s 完成保持验收。因此节点从视频 9.84 s 达成；视频 5.88 s 仅是瞬时位姿开始合格，不提前计为整步完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.00–2.40 s，单只 MyoHand 的拇指侧与屈曲其余手指围住、拨转实骰，骰子在掌指间改变朝向；视频 5.88–9.84 s 继续以掌指接触约束骰子，未见松手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better"
+  },
+  {
+   "name": "rot_dist",
+   "value": null,
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better"
+  },
+  {
+   "name": "solved_run",
+   "value": 100,
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "MyoHand 在掌指间拨转并约束骰子，视频 5.88 s 成功指示变绿，连续保持至视频 9.84 s 完成 100 控制步验收，最终达到唯一节点（1/1），评分器确定性重放成功。视频按每控制步一帧、25 fps 慢放，保持窗口在视频中跨越 3.96 s，不能按仿真 1.0 s 直接计时。没有状态日志，评分器 reward 未列出终态位置和朝向误差，故对应原生指标留空；solved_run=100 来自 log step 179 对相同交付动作的服务重放结果，判定结合视频绿色指示及最终评分。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "唯一节点 die_at_target_pose 从视频 9.84 s 达成（视频 5.88 s 指示变绿、按每控制步一帧换算 100 步验收），与帧和评分器结果一致；无失败层，时钟换算有变绿事件核对。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:52:24Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--khafnxf/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--khafnxf/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f6b9ea81f1040c84ec04a5cd0ee686b1baa12d7bc33919f198ca690203d3a1fb",
+   "video_sha256": "0465a996ada8589134f51b8bc2d8ff762c45f39b5e78b20a112b1ec054eaea93",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--zc4njd9.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--zc4njd9.json
new file mode 100644
index 0000000000000000000000000000000000000000..51855dbfa657dce19918512332142ef220a62990
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--zc4njd9.json
@@ -0,0 +1,201 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-privileged--zc4njd9",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 200,
+   "sim_steps": 200,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 141,
+   "agent_resets": 13442,
+   "agent_sim_steps": 85449,
+   "agent_records": 12011,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.47,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.26,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-privileged--zc4njd9/replay.mp4",
+  "duration_s": 8.04,
+  "fps": 25.0,
+  "frames": 201,
+  "sha256": "123af8306c02a59562058287f154965ac35c9565dc48695555bb2ee6182cfc2d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 实骰已在 MyoHand 掌指间，但彩色面朝向与旁侧半透明目标骰不同，目标外圈为红色，初态未达成。"
+     },
+     {
+      "t": 1.4,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–2.00 s 的密集帧显示 MyoHand 前臂及手腕转动、手指弯曲并带动实骰转向；视频 1.40 s 实骰仍在掌指间，目标外圈仍红。此后至视频 8.00 s 的全程抽帧以及视频 7.00–8.00 s 逐帧均未见绿色成功指示，末帧实骰仍留在手中。"
+     },
+     {
+      "t": 8.0,
+      "log_step": 175,
+      "text_zh": "观察：log step 174 重新生成并保存最终 200 步轨迹，服务终态 success=False、solved_run=0；log step 175 对该文件两次重放均 success=false，哈希相同。run.json 的评分器字段 replay_success=0、deterministic=1，与视频末段红色指示一致。"
+     },
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "推断：参考视频 7.20 s 可见相同视图中的目标外圈为绿色；本次缺少该可见线索，且已验证的交付未通过连续保持验收，故节点没有达成区间。旁侧目标模型有显示偏移，不以两骰模型是否重合判断位置误差。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.00–2.00 s，单只 MyoHand 的掌部和弯曲手指接触、约束实骰，前臂及手腕同时转动，实骰在掌指间改变朝向；视频 7.00–8.00 s 实骰仍被弯曲手指围住，未见松手。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": 1.4,
+   "log_step": 169,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 1.40 s，MyoHand 已带动骰子转向并弯指围住骰子，但目标外圈仍红；后续直到视频 8.00 s 未出现成功参考视频所示的绿色验收指示。log step 169 的交付脚本执行旋前脉冲后改用固定保持动作，log step 174–175 确认该 200 步交付及其重放均未成功。"
+  },
+  "unmet": {
+   "die_at_target_pose": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.4,
+    "log_step": null,
+    "evidence_zh": "观察：视频 1.40 s 及视频 7.00–8.00 s，实骰随手转动后仍停在掌指间，目标外圈持续为红色；没有出现参考视频 7.20 s 的绿色外圈。推断：重定向后的位姿未满足位置、朝向与连续保持的联合验收，未见脱手下落，故标位姿不准。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 169,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 169 的最终脚本只执行两段固定肌肉激活动作（35 步旋前脉冲、165 步保持），没有根据位置或朝向误差修正动作。同参数会话诊断 log step 164 显示朝向误差可降至 0.2359 rad，而位置误差仍为 0.0844 m、solved_run=0；这些诊断只作根因证据。log step 174–175 的实际交付和重放也均失败，视频 1.40–8.00 s 可见持骰后始终未进入绿色验收状态，支持控制未能同时校正位置与朝向；不足以确定具体控制子类。"
+   }
+  ],
+  "gap_zh": "没有状态日志，评分器 reward 未报告 pos_dist、rot_dist 或 solved_run 的数值，三个原生指标留 null。视频可确认红色指示及持骰状态，但不能独立测量精确位姿误差或连续计数；会话诊断不作为本次视频的逐帧状态，控制根因仅标到大类。"
+ },
+ "summary_zh": "MyoHand 已实际接触并带动掌中骰子转向，随后弯指持骰，但“骰子转到目标”始终未通过联合验收，最终为 0/1。断点在重定向操作中，现象为位姿不准；固定两段肌肉控制未能同时校正位置与朝向，最终文件两次确定性重放均失败。没有状态日志或评分器误差明细，不能给出精确误差与更细控制归因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点、位姿不准现象与控制根因均有视频与 log step 169/174/175 支持，未发现需要改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:51:24Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--zc4njd9/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--zc4njd9/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "29d4b01f9c468cb5f860cc3aae82479ad817d7f18eedebba377d43f47503fe58",
+   "video_sha256": "123af8306c02a59562058287f154965ac35c9565dc48695555bb2ee6182cfc2d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--zhftbwl.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--zhftbwl.json
new file mode 100644
index 0000000000000000000000000000000000000000..40b6564d4116ebac3fc028d00af7f978a0900752
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-privileged--zhftbwl.json
@@ -0,0 +1,176 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-privileged--zhftbwl",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 181,
+   "sim_steps": 181,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 60,
+   "agent_resets": 69,
+   "agent_sim_steps": 2081,
+   "agent_records": 0,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.84,
+   "peak_processes": 16,
+   "peak_disk_gb": 0.243,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-privileged--zhftbwl/replay.mp4",
+  "duration_s": 7.28,
+  "fps": 25.0,
+  "frames": 182,
+  "sha256": "666667734ceb1d7987e477faab38f894b934f4187ce7ca9515a8bb0dcb4354ca"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [
+     [
+      7.24,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的俯视与侧视中，实骰已位于 MyoHand 掌指间，彩色面及棱边朝向与旁侧半透明目标骰子明显不同，目标旁指示为红色，节点初始不成立。"
+     },
+     {
+      "t": 3.28,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–3.24 s，MyoHand 手指屈伸并配合腕部转动，使掌指间实骰改变朝向；视频 3.24 s 指示仍红，视频 3.28 s 变绿，实骰的彩色面朝向接近显示目标。视频 4.00、5.00、6.00 s 及逐帧检查的视频 6.20–7.24 s 均显示实骰仍在掌指间、指示为绿，未见脱手下落。"
+     },
+     {
+      "t": 7.24,
+      "log_step": 39,
+      "text_zh": "观察：log step 39 的基准 Session.replay 返回 success=True、steps=181、solved_run=100，并将该次重放截取为 181 行交付；同一步的评分器两次新进程重放均 success=true、终态哈希相同，log step 41 再次验证同一交付，两次重放仍成功；本次 reward 的 replay_success=1、deterministic=1、sim_steps=181。"
+     },
+     {
+      "t": 7.24,
+      "log_step": 39,
+      "text_zh": "推断：本次视频共 182 帧、25 fps，与初态加 181 个交付控制步相符，每控制步对应一视频帧；完成第 181 步对应视频 181/25=7.24 s。末尾连续 100 步应从第 82 步开始，对应视频 82/25=3.28 s，已用视频 3.24–3.28 s 的红转绿核对；因此完整保持判据在视频 7.24 s 才达成，绿色刚出现时尚未完成节点。100 个 10 ms 控制步的首末采样跨度为仿真 0.99 s，对应本次视频 3.96 s，不能按视频 1 s 计。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.00–3.28 s，单只 MyoHand 的手指围住掌指间实骰，通过屈伸并配合腕部转动拨转骰子；视频 3.28–7.24 s，手指继续约束和微调实骰，骰子一直留在手内。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": 100
+  }
+ ],
+ "failure": null,
+ "summary_zh": "本次评分轨迹完成“骰子转到目标”节点，最终 1/1：MyoHand 在掌指间拨转实骰，视频 3.28 s 指示转绿，并在视频 7.24 s 完成连续 100 个控制步的保持。log step 39、41 的交付重放成功且终态一致，本次评分器也确认成功，因此无失败断点。没有状态日志；位置和欧拉角误差的精确终值未由评分器字段单独报告，原生指标 pos_dist、rot_dist 留空，solved_run=100 来自 log step 39 对同一交付的基准服务重放结果。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功运行，单节点 die_at_target_pose 在视频 7.24 s（控制步 181/25，已用 3.24–3.28 s 红转绿核对）达成，final 为真，无失败断点，接触标注与证据一致。未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:54:02Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--zhftbwl/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-privileged--zhftbwl/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3202803ce908d2cecd976e577b80f35576de841dbd9a96774f53fc255d1356e4",
+   "video_sha256": "666667734ceb1d7987e477faab38f894b934f4187ce7ca9515a8bb0dcb4354ca",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--23whptx.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--23whptx.json
new file mode 100644
index 0000000000000000000000000000000000000000..e8ceae4d5d0fa617ca5657ef577ab6ff6f6dd596
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--23whptx.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-standard--23whptx",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 1375,
+   "sim_steps": 1375,
+   "physics_steps": 6875,
+   "steps": 4692,
+   "calls": 1041,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 4692,
+   "elapsed_s": 1952.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.487,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.749,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-standard--23whptx/replay.mp4",
+  "duration_s": 55.04,
+  "fps": 25.0,
+  "frames": 1376,
+  "sha256": "0a4f22a916d7d9aaad30dfb172209f0dbcc880465c6df42d519f9db127ba3ff2"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 实骰已在 MyoHand 掌指间，但彩色面朝向与旁侧半透明目标骰不同，目标旁指示呈红色，初态未达成。"
+     },
+     {
+      "t": 6.2,
+      "log_step": 58,
+      "text_zh": "观察：视频 4.4–8.4 s 的密集帧显示，MyoHand 拨动实骰，实骰在掌指间翻滚，随后仍露出与目标不同的彩色面；全片逐秒抽帧未见成功参考视频约 6.0–9.0 s 那样的持续绿色指示。log step 58 记录末回合重置为 episode 11，脚本输出 tipped at 155；这里只用视频帧确定动作时刻，不把 agent 时间换算为视频时间。"
+     },
+     {
+      "t": 54.6,
+      "log_step": 60,
+      "text_zh": "观察：视频 36.0–53.0 s 实骰逐渐向指端外移；视频 54.4–55.0 s 实骰脱离手指并远离手部，末帧未恢复。log step 60 中基准服务返回 episode=11、control_steps=1375、done=True、reason=failure: die dropped；评分器 live_success=0、replay_success=0、live_matches_replay=1，与视频末段一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频 4.4–8.4 s，MyoHand 单手拇指侧拨动实骰，掌面及其余手指承托骰子，骰子在手中滚转；后续拨动时逐渐移向指端。未见合手夹紧骰子的对向抓握。log step 58 的末回合脚本以 OP、FPL 拇指肌肉命令实施拨推。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": 54.6,
+   "log_step": 60,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 54.4–55.0 s，MyoHand 正在掌中调整骰子，实骰由指端脱离并远离手部，未再承接回来；目标旁指示仍为红色。log step 60 的末回合服务状态返回 failure: die dropped，整步未完成。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 54.6,
+    "log_step": 60,
+    "evidence_zh": "观察：视频 8.0–52.0 s 实骰一直由掌指承托，随后向指端移出；视频 54.6 s 已可见实骰与手部之间出现间隙，视频 54.8–55.0 s 继续远离。log step 60 的基准服务以 failure: die dropped 终止末回合，属于已有承托接触后失去物体。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 58,
+    "t": 54.6,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 58 的末回合脚本在拨转后重复固定的 80 控制步拇指拨推和 150 控制步等待，每轮自动将拨推幅度增加 0.15；estimate_now 虽计算骰子位姿误差，其结果仅打印并作为下次估计初值，没有用于修正拨推方向、幅度或承托姿态。视频 36.0–53.0 s 可见实骰向指端外移，视频 54.6–55.0 s 脱离手部；该控制循环没有把骰子留在掌中并纠正到目标。现有证据不足以进一步区分接触布局与反馈跟踪问题。"
+   }
+  ],
+  "gap_zh": "没有模拟器状态日志，评分器未报告 pos_dist、rot_dist 或 solved_run 的数值；log step 58、60 的误差数值来自 agent 自己的视觉拟合，不能代替评分器指标。视频可见目标指示未持续变绿和最终脱手，足以判定节点未达成，但不能精确恢复每个控制步的误差或确定具体肌肉力与接触失稳机制。",
+  "unmet": {
+   "die_at_target_pose": false
+  }
+ },
+ "summary_zh": "末回合中，MyoHand 拨转了掌中的骰子，但未达到并连续保持目标位姿，最终节点为 0/1。骰子逐渐移向指端，并在视频约 54.6 s 脱手；脚本周期性估计位姿，却仍按固定循环增加拨推幅度，未能纠正外移。没有状态日志或评分器误差读数，具体接触失稳机制无法细分。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点 die_at_target_pose 未达成、breakpoint 为该节点并在末回合脱手，与视频帧和服务返回 failure: die dropped 一致；现象“没保持住”和根因“控制”均有证据，且推断已标注。未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:56:36Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--23whptx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--23whptx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "58703a9aa9a2d97b60b52cc23704459e0221f86ead284b7dfdd280ed4f363646",
+   "video_sha256": "0a4f22a916d7d9aaad30dfb172209f0dbcc880465c6df42d519f9db127ba3ff2",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--63idwxz.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--63idwxz.json
new file mode 100644
index 0000000000000000000000000000000000000000..5ff7d2413a6a2899432245fca5608e1957669e5e
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--63idwxz.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-standard--63idwxz",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 114,
+   "sim_steps": 114,
+   "physics_steps": 570,
+   "steps": 504,
+   "calls": 51,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 504,
+   "elapsed_s": 3764.3,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.782,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.905,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-standard--63idwxz/replay.mp4",
+  "duration_s": 4.6,
+  "fps": 25.0,
+  "frames": 115,
+  "sha256": "956dee314748106b5cab4137065595bcb2427a868265a2a6ebf7676bef77bbdf"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 97,
+      "text_zh": "观察：视频 0.0 s，实骰已由 MyoHand 掌指承托，但彩色面和棱边朝向与旁侧半透明目标不同，目标周围为红色指示。log step 97 的服务返回 episode=11、resets_left=0、control_steps=0；任务文本规定只评分最后实时回合，此前重置的试验不计入本段。"
+     },
+     {
+      "t": 1.8,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0–3.4 s，MyoHand 前臂及腕部转动，实骰随手移动和转向；视频 1.8 s 实骰的红色单点面近乎正对侧视相机，目标对应面及棱边仍倾斜。逐段查看的视频中目标指示保持红色，未出现成功参考视频 5.6–9.84 s 可见的绿色指示。参考仅用于辨认成功线索。"
+     },
+     {
+      "t": 4.56,
+      "log_step": 113,
+      "text_zh": "观察：视频 3.52–4.56 s 逐帧查看，MyoHand 末段屈指使实骰继续倾转，但最终实骰红色单点面及相邻面的朝向仍与目标不同，目标周围仍为红色指示。log step 113 发出屈指肌肉批次；评分器 reward.live_success=0、task_failure=0、timeout=1，视频中没有达到并保持目标的片段。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.0–4.56 s，单只 MyoHand 的掌面及掌指承托实骰，前臂和腕部转动带动实骰改变姿态；视频 4.2–4.56 s 手指屈曲进一步拨转、约束实骰，骰子始终留在该手掌指附近。未见松手放置。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": 1.8,
+   "log_step": 98,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 1.8 s，MyoHand 已转腕并带动掌中实骰，但实骰与目标的面和棱边朝向仍不一致、指示为红；随后视频 2.0–4.56 s 前臂继续旋转并屈指调整，至末帧仍未修复。log step 98 开始最后回合的转腕、短旋前和保持动作。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 1.8–3.4 s，实骰红色单点面近乎正对侧视相机，而目标骰子的对应面和棱边明显倾斜；视频 4.56 s 手指虽已屈曲，实骰仍未取得目标朝向，目标指示保持红色。此判断依据朝向及成功指示，不把两骰显示中心的间距当作位置误差。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 98 的实际执行代码以固定肌肉批次依次施加转腕、4 行旋前及20 行保持命令，log step 102、107 再调整腕部肌肉，log step 109、111、113 分别调整拇指、伸指及屈指；服务返回的 pro_sup 从 log step 102 的 -1.008 rad 继续变至 log step 107 的 -0.854 rad 和 log step 113 的 -0.713 rad。结合视频 1.8–4.56 s 实骰持续随手转动、末段屈指后仍与目标朝向不合，肌肉动作未能把骰子稳定调到所需位姿；证据仅支持控制这一层，无法确定具体控制子类。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有状态日志；运行记录未报告评分器 pos_dist、rot_dist 或 solved_run 的数值，不能量化位置与欧拉角误差或精确恢复连续合格计数。节点未达由可见朝向差异、始终未见绿色成功指示及实时失败判定共同支持；根因无法细分至具体肌肉参数、接触设计或反馈机制。",
+  "unmet": {
+   "die_at_target_pose": false
+  }
+ },
+ "summary_zh": "最后实时回合中，MyoHand 转动前臂和腕部，并在末段屈指调整掌中骰子，但“骰子转到目标”始终未达成，最终为 0/1。视频 1.8–4.56 s 的实骰朝向持续偏离目标且成功指示保持红色，控制动作未能实现稳定目标位姿；缺少精确误差与连续合格计数，无法细分控制根因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "单节点任务，节点未达、断点、位姿不准现象与控制根因均有视频和日志支持；末帧骰子朝向仍与目标不同且指示为红，未发现需改动之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:55:27Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--63idwxz/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--63idwxz/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c0d076054545f2b7069c0796c7fa2df15e63585a50112ef4aa4c3974cc82da3c",
+   "video_sha256": "956dee314748106b5cab4137065595bcb2427a868265a2a6ebf7676bef77bbdf",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--72arrhi.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--72arrhi.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d1c19ca9f89e1a0793d50aff62b27e19679ba42
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--72arrhi.json
@@ -0,0 +1,182 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-standard--72arrhi",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 1577,
+   "sim_steps": 1577,
+   "physics_steps": 7885,
+   "steps": 3674,
+   "calls": 2711,
+   "episodes": 5,
+   "resets": 4,
+   "reset_budget": 10,
+   "sim_steps_total": 3674,
+   "elapsed_s": 3006.9,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-standard--72arrhi/replay.mp4",
+  "duration_s": 31.6,
+  "fps": 25.0,
+  "frames": 790,
+  "sha256": "d672f42189767aef931dcc0dc0c303dcd627a8eec2d59780469d049545eaf795"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [
+     [
+      31.56,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 实骰已在 MyoHand 掌指间，实骰与旁侧半透明目标骰子的彩色面朝向不同，目标旁指示为红色；初态尚未完成重定向。"
+     },
+     {
+      "t": 8.0,
+      "log_step": null,
+      "text_zh": "观察：视频 2.0–9.0 s 单只 MyoHand 的手指围住实骰，前臂和腕部转动时实骰随之改变朝向；视频 24.0–29.4 s 手指逐渐调整弯曲程度，实骰仍留在掌指间。"
+     },
+     {
+      "t": 29.6,
+      "log_step": null,
+      "text_zh": "观察：视频 29.56 s 目标旁指示仍红，视频 29.60 s 已变绿；检查视频 29.6–31.4 s 的连续取样及视频 31.36–31.56 s 的逐帧图，绿色保持，实骰未脱手，末帧彩色面与目标朝向接近。此处变绿只表明即时位姿进入容差，尚不能立即记为完成连续保持。"
+     },
+     {
+      "t": 31.56,
+      "log_step": 133,
+      "text_zh": "观察：log step 133 的机器人服务返回 episode=5、control_steps=1577、sim_time_s=15.77、done=true、reason=success；评分器 reward 同时给出 live_success=1、replay_success=1、deterministic=1、live_matches_replay=1。视频 31.56 s 末帧仍显示绿色指示且实骰在手中。"
+     },
+     {
+      "t": 31.56,
+      "log_step": 134,
+      "text_zh": "推断：log step 92 重置后开始最终回合，log step 134 保存的成功序列合计 1577 个控制步，与评分器 n_actions=1577 及视频 790 帧一致。结合控制周期 0.01 s、视频 25 fps 和末帧视频 31.56 s，推断本视频约为仿真时钟的 2 倍；连续 100 步约占视频 2 s，与视频 29.60 s 变绿后保持至视频 31.56 s 的终止吻合。因此以已检查的末帧视频 31.56 s 作为联合判据首次完成的锚点，精度约 0.1 s；未把 agent 工作时间或参考视频的慢放比例用于换算。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 2.0–9.0 s 及视频 24.0–31.56 s，单只 MyoHand 的拇指和其余手指围住、约束实骰，配合腕部与前臂转动，并调整手指弯曲程度让骰子在手中改变朝向和位置；没有把实骰松开放到外部支承面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "最终受评的是第 5 回合：MyoHand 在掌指间转动并调整骰子，视频 29.60 s 指示变绿，至视频约 31.56 s 满足连续保持验收，唯一节点完成，终态 1/1；服务及重放评分均确认成功。此前重置掉的回合不计入该视频进度，成功轨迹没有失败断点。没有状态日志，评分器也未报告 pos_dist、rot_dist、solved_run 的具体值，故原生指标留空；保持完成时刻由末帧、服务终止状态及控制步数共同确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "唯一节点 die_at_target_pose 在视频约 31.56 s 达成，与指示灯 29.60 s 变绿后保持 100 控制步（视频约 2 s）及评分器成功一致；时钟换算有依据，接触标注合理，无失败断点。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:56:49Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--72arrhi/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--72arrhi/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b6f30cf0172cb6f912150882179c6a60b80b643a78a29216f50eced7b620c7db",
+   "video_sha256": "d672f42189767aef931dcc0dc0c303dcd627a8eec2d59780469d049545eaf795",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--em2sjvc.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--em2sjvc.json
new file mode 100644
index 0000000000000000000000000000000000000000..4ec23beff7180d2056b82863b9e0cab9b6090667
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--em2sjvc.json
@@ -0,0 +1,222 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-standard--em2sjvc",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 50,
+   "sim_steps": 50,
+   "physics_steps": 250,
+   "steps": 103,
+   "calls": 4,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 103,
+   "elapsed_s": 2805.4,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.634,
+   "peak_processes": 14,
+   "peak_disk_gb": 0.464,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-standard--em2sjvc/replay.mp4",
+  "duration_s": 2.04,
+  "fps": 25.0,
+  "frames": 51,
+  "sha256": "206062c88e04f8784bc245aaeb1e7f8e442619319554073a5cd89824007c0660"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 65,
+      "text_zh": "观察：log step 65 重置服务并返回 episode=2；任务文本规定只评分最后回合。视频 0.00 s 实骰已在 MyoHand 掌指处，实骰与旁侧半透明目标骰子的彩色面、棱边朝向不同，目标旁指示为红色。"
+     },
+     {
+      "t": 1.4,
+      "log_step": 67,
+      "text_zh": "观察：已逐帧检查视频 0.00–2.00 s 的全部 51 帧，包括视频 1.04–2.00 s 末段。MyoHand 腕部弯曲、手指随之收拢，实骰在掌指旁转动和移位，但朝向一直不同于目标，指示始终未变绿；未见目标位姿及连续保持成立。"
+     },
+     {
+      "t": 2.0,
+      "log_step": 74,
+      "text_zh": "观察：log step 67 的脚本向第 2 回合提交 50×39 的全零肌肉指令；log step 74 服务返回 episode=2、control_steps=50、done=False、reason=None。评分字段 live_success=0、n_actions=50、task_failure=0；视频末帧的实骰仍与目标朝向不符。本回合没有 log step 64 所述第 1 回合的掉落终止。"
+     },
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：成功参考 inputs/oracle/replay.mp4 的参考视频约 6.5–9.84 s 出现绿色指示且骰子留在掌指间；本次视频全段均无该提示。目标骰子含显示偏移，此处比较彩色面朝向与成功指示，不以两骰画面中心是否重合作为判据。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": null,
+   "log_step": 67,
+   "where": "首步前",
+   "evidence_zh": "观察：log step 65 重置后进入评分的第 2 回合；log step 67 为比较服务与 agent 自建模型，仅提交 50×39 的全零肌肉指令。视频 0.00–2.00 s 全部 51 帧可见腕指弯曲及实骰移位，未见朝目标的主动重定向操作；log step 73 只写出控制脚本并作语法检查，没有运行其中的优化与回放，log step 74 服务仍为 episode=2、control_steps=50、done=False。因此目标步骤从未启动，按 docs/03_trajectory.md Step 3.1 及 CASES.md FAIL-13 标为首步前，不把 agent 的会话测试标成机器人探索；没有可定位的失败瞬间，t 留空。"
+  },
+  "unmet": {
+   "die_at_target_pose": false
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": null,
+    "log_step": 67,
+    "evidence_zh": "观察：评分第 2 回合只在 log step 67 提交了 50×39 的全零肌肉指令，用于服务与本地模型的状态比对；视频 0.00–2.00 s 可见腕指弯曲和实骰随之移位，未见朝目标的主动操作。log step 73 仅写出脚本并检查语法，log step 74 的服务读数仍为 episode=2、control_steps=50，目标重定向步骤未启动（docs/03_trajectory.md Step 3.1–3.2；CASES.md FAIL-13）。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "观察：inputs 中无状态日志，评分字段未提供 pos_dist、rot_dist、solved_run，故三项原生指标均为 null，无法恢复精确误差和连续计数。评分回合只有零激励会话测试，未有对目标步骤的控制尝试；现象标为“没尝试”，不将被动移位后的姿态标为“位姿不准”，也不从第 1 回合借用掉落现象（CASES.md FAIL-13、FAIL-23）。log step 67 的模型对比和 log steps 69–72 的离线拟合未验证服务与本地模型的差异来源，也未显示可行目标方案的控制执行失败；agent 在 log steps 73、77 自述剩余时间及模型差异是其停止理由，不能直接当作机器人失败的根因。会话证据虽可辅助归因，但此处不足以确定原因，按 docs/03_trajectory.md Step 3.2–3.3、TAXONOMY_v2.md 规则 6 将根因留空。具体手指接触对及后段是否持续承托仍缺少直接测量。"
+ },
+ "summary_zh": "评分的第 2 回合只执行了 50 步全零肌肉指令的会话测试，骰子随腕指运动而移位，但视频全段朝向与目标不同、成功指示未变绿，“骰子转到目标”最终为 0/1。目标步骤从未启动，断点为“首步前”，现象为“没尝试”；后续优化及重规划控制未实际执行，前一回合的掉落不属于本次评分轨迹。没有状态日志或精确位姿误差，本地模型对比和 agent 的停止说明也不足以确定根因，故根因留空。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点值 0/1 与视频一致（全段指示未变绿、朝向不符）。但断点 where 与根因有误：评分回合只有 agent 为核对自建孪生模型发出的 50 步零激励会话探测，没有任何朝目标的动作，按 CASES.md FAIL-13 应为「首步前」+ 现象「没尝试」，不是「探索」；根因「控制」仅来自 agent 自述的本地孪生与服务差异（会话测试），不是评分回合的证据，且不能从没尝试反推根因。",
+    "n_issues": 3,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点 0/1、断点“首步前”、现象“没尝试”、根因留空均与视频和 log 一致；第 1 轮意见已正确落实。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:35Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--em2sjvc/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--em2sjvc/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--em2sjvc/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--em2sjvc/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3745582c63062366a79b0d9f03127da6933d92f5a2235f52052b71440825dc11",
+   "video_sha256": "206062c88e04f8784bc245aaeb1e7f8e442619319554073a5cd89824007c0660",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--fjqqusm.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--fjqqusm.json
new file mode 100644
index 0000000000000000000000000000000000000000..9511832163d024c674d0ee0f60383b7c49b500b9
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--fjqqusm.json
@@ -0,0 +1,208 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-standard--fjqqusm",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 30,
+   "sim_steps": 30,
+   "physics_steps": 150,
+   "steps": 549,
+   "calls": 71,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 549,
+   "elapsed_s": 3649.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.909,
+   "peak_processes": 12,
+   "peak_disk_gb": 1.08,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-standard--fjqqusm/replay.mp4",
+  "duration_s": 1.24,
+  "fps": 25.0,
+  "frames": 31,
+  "sha256": "a6681e88ccf0a9580c3a88b0f2f3038fb371a0542a5531954f165cef20b8cc22"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 实骰位于 MyoHand 掌指间，彩色面朝向与旁侧半透明目标不同，目标周围指示为粉红色；逐帧检查视频 0.00–1.20 s，指示未变绿。成功参考末段同一显示会变绿，旁侧目标具有显示偏移，未用两骰中心重合来判断。"
+     },
+     {
+      "t": 0.96,
+      "log_step": 210,
+      "text_zh": "观察：视频 0.48–0.96 s 手腕和手指弯曲，实骰由掌指间移到手指外侧；视频 1.04–1.20 s 实骰进一步离开手部，未恢复目标位姿。log step 210 的服务状态为最后回合 episode=11、control_steps=30、done=true、reason=\"failure: die dropped\"；reward 的 live_matches_replay=1，说明该末回合就是评分录像所示回合。"
+     },
+     {
+      "t": 1.2,
+      "log_step": 210,
+      "text_zh": "推断：该回合只有 30 个控制步，且全片无绿色成功指示，不能完成节点要求的连续 100 步位姿合格保持，因此此节点从未成立。31 帧、25 fps 的录像是慢放，未把视频 1.24 s 当作仿真 1.24 s，也未用 agent 自算的位置与朝向误差替代评分器指标。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频 0.00 s MyoHand 单手掌部从下方托着实骰；视频 0.24–0.72 s 手腕转动、手指弯曲时仍与实骰接触，随后实骰移出掌指承托区域，视频 1.20 s 已与手部分离。未见能持续约束骰子的对向夹持，不把手指弯曲等同于成功抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": 0.96,
+   "log_step": 210,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.72–1.20 s，正在调姿的 MyoHand 未将实骰留在掌指间，实骰逐渐越出手指外侧并与手部分离，指示保持粉红色；log step 210 的最后回合服务返回 failure: die dropped，之后未修复。视频 0.96 s 为失持过程的粗略导航点。"
+  },
+  "unmet": {
+   "die_at_target_pose": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 0.96,
+    "log_step": 210,
+    "evidence_zh": "观察：视频 0.00 s 实骰受单手掌指承托；视频 0.72–1.20 s 手指弯曲、腕部转动时，实骰移至手指外侧，末帧已与手部分离。log step 210 的服务终止原因为 failure: die dropped；这是调姿中失去原有承托，未见目标内计划松手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 210 的实际末回合脚本每次执行 6 个控制步，随后对腕部及指部肌肉激活作随机扰动；输出前两次为 nomeas，代码在测量失败时仍令 a=a_try 并继续扰动，测量成功后也未按误差方向校正或拒绝较差指令。结合视频 0.24–1.20 s 的腕指弯曲及实骰逐步脱离，指令没有维持住承托并完成调姿；证据支持控制失效，但不足以区分接触几何、力大小或肌肉动态的具体贡献。"
+   }
+  ],
+  "gap_zh": "没有状态日志，无法核验各手指接触对、接触力、逐步位置及欧拉角误差或连续合格计数。评分器未报告 pos_dist、rot_dist、solved_run，native_metrics 均置 null；log step 210 的 dp/da 是 agent 自己的视觉估计，且采用旁侧显示目标中心，不能作为这些原生指标。控制子类型尚不能可靠细分。"
+ },
+ "summary_zh": "评分录像为最后一次实时回合，唯一节点“骰子转到目标”始终未达成，终态为 0/1；手腕与手指运动时，原受承托的实骰移出掌指区域并在末段脱离，服务判定掉落。最后回合的脚本在位姿测量缺失时仍继续随机扰动肌肉指令，根因保守归为控制；无状态日志和评分器误差序列，不能进一步核验接触力及控制子类型。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、终态 0/1、断点、接触（托/舀）、现象“没保持住”与根因“控制”均有视频和 log step 210 的服务读数支持；时钟均标明，未见 agent 时间被换算成视频时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:49:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--fjqqusm/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--fjqqusm/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9a761e3db421f020253db5efa0c9bf58dcfedab1b4940f9778b6e8e019ec6b77",
+   "video_sha256": "a6681e88ccf0a9580c3a88b0f2f3038fb371a0542a5531954f165cef20b8cc22",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--two37kb.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--two37kb.json
new file mode 100644
index 0000000000000000000000000000000000000000..434fcbfdf5933809091c5217f064fc288c7142d0
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--two37kb.json
@@ -0,0 +1,213 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-standard--two37kb",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 315,
+   "sim_steps": 315,
+   "physics_steps": 1575,
+   "steps": 1408,
+   "calls": 119,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 1418,
+   "elapsed_s": 799.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.222,
+   "peak_processes": 6,
+   "peak_disk_gb": 0.252,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-standard--two37kb/replay.mp4",
+  "duration_s": 12.64,
+  "fps": 25.0,
+  "frames": 316,
+  "sha256": "4d7f8b210194b72ef4b80c46533d9a704df183f661c18aa5e635136bfa96e8c0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 实骰已在 MyoHand 掌指间，但实骰与旁侧半透明目标骰子的彩色面朝向不同，目标外圈为红色，初态未达成。"
+     },
+     {
+      "t": 5.0,
+      "log_step": null,
+      "text_zh": "观察：已检查视频 0.0–12.5 s 每约 0.5 s 的双视图、视频 0.0–3.8 s 每约 0.2 s 的动作及视频 11.4–12.6 s 密集末段。手指卷曲、实骰翻转，手腕明显转动；实骰未保持目标位姿，目标外圈始终未见参考成功视频末段那样的绿色。旁侧模型有显示偏移，不以两骰中心是否重合判定。"
+     },
+     {
+      "t": 12.6,
+      "log_step": 120,
+      "text_zh": "观察：视频 12.2–12.6 s 实骰向掌侧缘滑出，在左视图中从手部轮廓旁显露；目标指示仍为红色。log step 120 的机器人服务记录最后一次重置后为 episode 11，在 control_steps=315 返回 done=True、reason=failure: die dropped；reward 的 live_matches_replay=1、live_success=0、replay_success=0 与本段相符。"
+     },
+     {
+      "t": 12.6,
+      "log_step": 120,
+      "text_zh": "推断：评分回合是 log step 120 重置后的最后回合，前十回合仅作调试证据。本视频共 316 帧，含初帧及 315 个控制步，故控制步 k 对应视频 k/25 s；末次掉落返回 315/25=12.6 s，已与末帧滑出事件核对。100 个控制步需约 4 s 视频，不能将 1 s 仿真保持时间当作 1 s 视频；本次没有形成该达标保持段。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.0–3.8 s，MyoHand 单只手的拇指和其余手指在掌指间约束、拨转实骰，实骰随手指卷曲改变朝向；视频中后段继续由弯曲手指承托、约束，视频 12.2–12.6 s 向掌侧缘滑移。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": 12.2,
+   "log_step": 120,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 12.2–12.6 s，MyoHand 仍在卷指维持实骰，实骰却逐渐滑到掌侧缘之外，未恢复到目标位姿；log step 120 的最后评分回合随后以 failure: die dropped 结束。失败发生在掌中调姿及保持过程中。"
+  },
+  "unmet": {
+   "die_at_target_pose": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 12.2,
+    "log_step": 120,
+    "evidence_zh": "观察：视频 5.0 s 实骰仍被卷曲的手指围在掌指间；视频 11.8–12.6 s 的逐帧末段显示实骰逐渐从原先约束位置向手侧滑移，最后在左视图中突出手部轮廓，未被手指收回。log step 120 的机器人服务在最后回合 control_steps=315 明确返回 failure: die dropped；这是调姿时失去稳定持物，并非目标处的计划松手。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 119 的最终控制脚本先执行分段固定肌肉激活，随后重复同一个 20 控制步的 hold_chunk，仅检查服务的 done/reason，没有依据骰子位姿或腕姿修正激活；log step 120 证实该脚本在最后回合持续执行至掉落。视频 0.8–5.0 s 手腕带着实骰明显移位，视频 12.2–12.6 s 实骰继续滑出，说明这组控制未能同时稳定腕姿与掌中持物。现有证据不足以细分到接触布局、跟踪或具体参数。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有状态日志；run.json 的 reward 未报告 pos_dist、rot_dist 或 solved_run，三项原生指标均留 null，不能用 agent 在此前调试回合自行估算的误差替代评分值。末段手部骨骼遮挡了部分指面接触，无法确定最后失去的是哪一对手指接触及具体接触力，故根因只标控制，不细分。"
+ },
+ "summary_zh": "评分回合中，MyoHand 在掌指间拨转了骰子，但没有出现目标绿色成功指示，也未完成“骰子转到目标”的连续保持，最终节点为 0/1。视频 12.2–12.6 s 实骰向掌侧缘滑出，最后回合服务返回“failure: die dropped”；固定肌肉激活未能稳定调姿和持物，根因归为控制。无状态日志及评分器位姿误差、连续计数，无法进一步确定具体控制子因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、评分回合识别(episode 11,315 控制步=12.6 s 视频)、断点与「没保持住」均与帧和 log step 120 的服务读数一致;根因「控制」较笼统但无反证。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:52:48Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--two37kb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--two37kb/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9227aee62b0fe20ed450a2effa3df4fd11fdbb1c5772a60498b51e1a6c9775b1",
+   "video_sha256": "4d7f8b210194b72ef4b80c46533d9a704df183f661c18aa5e635136bfa96e8c0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--v4rqarh.json b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--v4rqarh.json
new file mode 100644
index 0000000000000000000000000000000000000000..81c5560ede36648ec05c020906da493fe725c255
--- /dev/null
+++ b/phase2/myosuite/die_reorient/trajectories/myosuite-die-reorient-i00-standard--v4rqarh.json
@@ -0,0 +1,201 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-die-reorient-i00-standard--v4rqarh",
+ "taskKey": "myosuite/die_reorient",
+ "dag_sha256": "8503244878c996302f5b3d2b56939acc4c42a7bd943ead0f8ea31be095176f02",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "physics_steps": 0,
+   "steps": 0,
+   "calls": 3,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 0,
+   "elapsed_s": 3641.2,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.155,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.262,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-die-reorient-i00-standard--v4rqarh/replay.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "e640165fe308d7271a006e8c3a32a564ff88c1c24569de36e2840d992e2a220e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "die_at_target_pose": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：已检查本次视频全部唯一一帧（视频 0.00 s，片长 0.04 s）；MyoHand 手掌旁的实骰彩色面和棱边朝向与半透明目标骰子不同，目标旁指示仍为红色。成功参考视频末段可见绿色指示，本次没有该达成线索。"
+     },
+     {
+      "t": 0.0,
+      "log_step": 4,
+      "text_zh": "观察：log step 4 中实时服务 status 的 control_steps=0、step=0；本次评分器最终字段 n_actions=0、sim_steps=0、physics_steps=0、resets=0、success=0、timeout=1，且 live_matches_replay=1。"
+     },
+     {
+      "t": 0.0,
+      "log_step": 74,
+      "text_zh": "推断：结合视频 0.00 s 的初态及评分器零动作、零控制步，评分回合从未启动掌中重定向，也没有连续 100 个合格控制步；log step 41、65、71、74 的运动代码均运行本地 MuJoCo 优化，没有提交到实时服务，不能计入本节点进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "观察：评分器记录 n_actions=0、sim_steps=0，视频仅有初态；agent 未向评分服务交付控制动作轨迹。初态骰子位于 MyoHand 掌指旁，不能将该初始布置或本地仿真运动记为本次实际手内调整。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "pos_dist",
+   "unit": "m",
+   "threshold": 0.025,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "rot_dist",
+   "unit": "rad",
+   "threshold": 0.262,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "die_at_target_pose",
+   "t": 0.0,
+   "log_step": 74,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0.00 s 是唯一初态帧，骰子尚未达到目标朝向，指示为红色；评分器最终 n_actions=0、sim_steps=0。log step 74 仍在启动本地轨迹优化，日志没有向实时服务提交运动的记录，评分回合的转骰步骤未开始。"
+  },
+  "unmet": {
+   "die_at_target_pose": false
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 74,
+    "evidence_zh": "观察：视频 0.00 s 仅显示未重定向的初态，评分器 n_actions=0、control 对应的 sim_steps=0、resets=0；log step 74 的脚本只创建本地 MyoSuite 环境并优化肌肉动作，没有调用实时 Robot.submit。评分回合没有启动所要求的转骰动作。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "没有状态日志，评分器也未报告 pos_dist、rot_dist 或 solved_run，不能填写精确原生指标。log step 68 只展示部分本地优化结果（角度误差仍超阈值且 consec_solved=0）；log step 71、74 的后续优化仅记录启动，未提供最终结果，也没有明确记录为何始终未提交实时动作，因而不足以确定控制、规划或交付根因，根因层留空。评分器 timeout=1、infra_error=0；预算结束本身不作为根因，本地优化中的数值警告也不证明评分环境异常。"
+ },
+ "summary_zh": "本次评分回合只有视频 0.00 s 的一帧初态，骰子朝向未对准且指示为红色；唯一节点“骰子转到目标”未达成，最终为 0/1。评分器记录零动作、零控制步，断点为该节点的首步前，现象为“没尝试”；日志中的转骰测试均属于本地仿真。最后几轮优化结果及未提交实时动作的明确原因没有记录，因此根因留空，三个原生指标也因评分器未报告而保留 null。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标注与证据一致：评分回合仅一帧初态，评分器 n_actions=0、sim_steps=0，run.json sim_total submit=0，全部转骰优化均为本地仿真；节点未达成、首步前、没尝试、根因留空均有依据（同 CASES FAIL-22）。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:52:16Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--v4rqarh/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-die-reorient-i00-standard--v4rqarh/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "47f5a97180c8daad2982ad00cd00a0002646c0b6fdeb2be59442aae4ade1e470",
+   "video_sha256": "e640165fe308d7271a006e8c3a32a564ff88c1c24569de36e2840d992e2a220e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--ebzqcpv.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--ebzqcpv.json
new file mode 100644
index 0000000000000000000000000000000000000000..f0c16fe842d1d07a5637b92042897da3d70f5f0f
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--ebzqcpv.json
@@ -0,0 +1,176 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-privileged--ebzqcpv",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 156,
+   "sim_steps": 156,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 62,
+   "agent_resets": 57643,
+   "agent_sim_steps": 94506,
+   "agent_records": 4,
+   "agent_replays": 194,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-privileged--ebzqcpv/replay.mp4",
+  "duration_s": 6.28,
+  "fps": 25.0,
+  "frames": 157,
+  "sha256": "a239c0d35560a0f0f6b2e1d1dcffae85d8466b5b986fabf701e1c7d75bf4a080"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [
+     [
+      6.24,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 7,
+      "text_zh": "观察：视频0.00 s黄色物体已在灵巧手的手指间，指示为红色；视频0.16–2.24 s手腕和手指运动、物体朝向改变，尚不能据此认定连续保持完成。log step 7的初态输出给出obj_err_pos=[0,0,0.005]，obj_rot与obj_des_rot不同。"
+     },
+     {
+      "t": 2.28,
+      "log_step": null,
+      "text_zh": "观察：逐帧查看视频1.80–2.60 s，指示在视频2.24 s仍为红色、视频2.28 s变绿；黄色物体仍在手指间。该颜色有锁存，单凭变绿不判定100步保持完成。"
+     },
+     {
+      "t": 6.24,
+      "log_step": 143,
+      "text_zh": "观察：视频全程以0.16 s间隔查看，并逐帧查看视频5.20–6.24 s，末段黄色物体持续位于手指间，未见脱手或明显转离；log step 141明确重新生成最终选定轨迹，log step 142记录该156行交付的两次新进程重放均成功、终态哈希相同，log step 143服务状态为steps=156、solved_run=100、hold_steps=100、success=True。inputs/run.json的评分字段也为success=1、replay_success=1、deterministic=1、n_actions=156。"
+     },
+     {
+      "t": 6.24,
+      "log_step": 143,
+      "text_zh": "推断：最终交付156个控制步对应视频157帧（含初始帧），25 fps下控制步k对应视频k/25 s，而非仿真k×0.01 s；由最终solved_run=100推得连续达标从控制步57开始，57/25=视频2.28 s，与逐帧看到的首次变绿核对一致。故控制步156完成保持窗口，节点从视频6.24 s达到并保持至结束；此前绿色阶段只是保持窗口进行中。无逐帧状态日志，精确阈值依赖该交付的服务状态及成功评分共同佐证。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.00–2.28 s，单只MyoHand的手指围住黄色棱角物体并改变接触姿态，配合手腕转动使物体转向；视频2.28–6.24 s仍将物体保持在手指间，未见释放或脱手。左右画面是同一只手的不同视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "单只灵巧手在手指间调整黄色物体朝向并持续保持，最终完成1/1节点，评分器确认交付轨迹成功且可确定性重放。视频约2.28 s首次变绿，结合最终156控制步、solved_run=100及逐帧映射，连续保持于视频6.24 s完成，无失败断点。没有逐帧状态日志，精确朝向和位置阈值由最终交付的服务记录及评分结果佐证；评分器reward未单列DAG中的三个原生指标，其value保留null。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，节点object_reoriented于视频6.24 s达成（控制步156，solved_run=100，57/25=2.28 s与变绿一致），无断点，接触标签合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:47:44Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--ebzqcpv/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--ebzqcpv/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "5d8cec136b7bb8e17ab4881522d6d3d23e11736fab5e8431f4e841d644e47ec3",
+   "video_sha256": "a239c0d35560a0f0f6b2e1d1dcffae85d8466b5b986fabf701e1c7d75bf4a080",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--fmo4i5g.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--fmo4i5g.json
new file mode 100644
index 0000000000000000000000000000000000000000..b7bec21cba45925ae3e5cefd1fb23932b460b448
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--fmo4i5g.json
@@ -0,0 +1,178 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-privileged--fmo4i5g",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 1050,
+   "sim_steps": 404,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 107,
+   "agent_resets": 297,
+   "agent_sim_steps": 18562,
+   "agent_records": 7,
+   "agent_replays": 27,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.298,
+   "peak_processes": 14,
+   "peak_disk_gb": 0.987,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-privileged--fmo4i5g/replay.mp4",
+  "duration_s": 16.2,
+  "fps": 25.0,
+  "frames": 405,
+  "sha256": "4da85b3b14fd37e615c3c43bf4b65dcd14e0bfbb55f76823fc245654e58d08dd"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [
+     [
+      16.16,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00 s黄色物体已在单只MyoHand的手指间，目标周围指示为红色；视频0.20–3.20 s灵巧手收拢手指并带动物体转动，物体没有离手。"
+     },
+     {
+      "t": 12.2,
+      "log_step": null,
+      "text_zh": "观察：视频6.00–12.16 s黄色物体仍夹在拇指与其余手指之间，手指及手腕继续调整其朝向；视频12.16 s指示仍红，视频12.20 s变绿。视频12.40、13.00、14.00、15.00及15.20–16.16 s的双视角帧显示物体继续留在手内，朝向基本稳定；末帧无掉落。"
+     },
+     {
+      "t": 16.16,
+      "log_step": 216,
+      "text_zh": "观察：log step 216最终重新生成所交付的1050行轨迹，基准Session报告steps=404、success=True、solved_run=100；同一步的基准重放器两次均报告success=true、steps=404、instance_ok=true、deterministic=true，终态哈希相同。inputs/run.json评分字段也给出success=1、replay_success=1、sim_steps=404。"
+     },
+     {
+      "t": 16.16,
+      "log_step": 216,
+      "text_zh": "推断：本次视频为25 fps、405帧，与最终交付重放的初帧加404个控制步一致，按每控制步一帧映射为视频t=控制步/25。视频12.20 s的第305帧开始变绿，至视频16.16 s的第404帧物体保持在手内，与终态solved_run=100的约4秒视频保持段相符；因此完成100步保持的节点首次记在视频16.16 s，而不在首次变绿时记完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.00–3.20 s单只MyoHand的手指围住黄色物体，随后在视频6.00–12.40 s通过拇指与其余手指的接触及手腕转动改变物体朝向；视频12.40–16.16 s物体持续位于手指间，未松手，保持同一手内的调姿结果。左右画面是同一只手的两个相机视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "观察：本次评分成功，单只灵巧手在手内调整黄色物体，视频12.20 s开始指示变绿，末段至视频16.16 s物体持续留在手指间并保持稳定。推断：结合最终交付重放在404个控制步结束、solved_run=100及405帧视频的对应关系，唯一节点“物体调至目标”在视频16.16 s完成，终态1/1，无失败断点。观察：本次没有逐帧状态日志，精确朝向余弦和位置误差不能由图像量出，连续保持由最终交付的基准结果支持；评分字段未单独报告DAG列出的三项原生指标，native_metrics值均留空。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "评分成功，唯一节点在视频16.16 s（控制步404，solved_run=100映射）记为达成，与视频12.2 s变绿、末帧绿色保持一致；无失败断点，contact_actual 合理。未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:43:59Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--fmo4i5g/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--fmo4i5g/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d247aa51f24caf1559f4120b6695ef7c4fd23a7327846c7549c1c4b89b1f21df",
+   "video_sha256": "4da85b3b14fd37e615c3c43bf4b65dcd14e0bfbb55f76823fc245654e58d08dd",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--gd33wrr.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--gd33wrr.json
new file mode 100644
index 0000000000000000000000000000000000000000..0ae0a800217eee2cf6b11e58b799e6069d1a7a41
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--gd33wrr.json
@@ -0,0 +1,197 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-privileged--gd33wrr",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 69,
+   "sim_steps": 69,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 86,
+   "agent_resets": 37964,
+   "agent_sim_steps": 14245,
+   "agent_records": 2,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 3.371,
+   "peak_processes": 21,
+   "peak_disk_gb": 0.249,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-privileged--gd33wrr/replay.mp4",
+  "duration_s": 2.8,
+  "fps": 25.0,
+  "frames": 70,
+  "sha256": "c971770ad71151fb3f1a46ce5f3886e0445a9f02b77ca1589be50251acddee9e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00–1.60 s的双视角显示，单只肌骨手的手指承托黄色方块，手腕逐渐转动；方块朝向随之改变，未出现稳定包握后的持续保持。"
+     },
+     {
+      "t": 2.08,
+      "log_step": 77,
+      "text_zh": "观察：视频2.08 s指示首次变绿，之后视频2.40–2.76 s方块逐渐与手指分离、远离手部；log step 77的交付生成脚本输出最大朝向余弦0.9978372855447364，同时记录和重放均在69个控制步报告failure: object dropped。绿色指示是瞬时达标后锁存，不能证明完成保持。"
+     },
+     {
+      "t": 2.76,
+      "log_step": 81,
+      "text_zh": "推断：log step 81明确重放最终交付文件，两次均为69行、69个控制步，success=false、reason=failure: object dropped且终态哈希相同，与视频2.76 s的方块脱离手部相符；整段交付不足所需的100个控制步，因此从未完成连续保持，节点全程为假。参考视频2.40–6.24 s可见方块持续留在收拢的手指间，本次没有该保持阶段。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频0.00–1.60 s单只MyoHand的手指从下方承托黄色方块，并随手腕转动带动方块滚转；视频2.40–2.76 s方块脱离承托。未见手指形成持续的对向包夹，不标抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_reoriented",
+   "t": 2.4,
+   "log_step": 81,
+   "where": "操作中",
+   "evidence_zh": "观察：视频2.40–2.76 s，肌骨手仍在转腕、伸屈手指，黄色方块却离开手指并持续远离，未被重新接住；log step 81对交付文件的两次独立重放均在69个控制步因object dropped终止。失败发生在物体调姿过程中，连续100步保持从未完成。"
+  },
+  "unmet": {
+   "object_reoriented": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.4,
+    "log_step": 81,
+    "evidence_zh": "观察：视频0.00–1.60 s黄色方块原由单手手指承托；视频2.40–2.76 s它与指尖间出现并扩大空隙，末帧已明显脱离手部，没有重新承接。log step 81的交付重放器同时报告failure: object dropped，说明本次调姿未维持物体的承托与控制。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 10、11、36中的脚本把关节关键帧经肌肉PD控制转换为动作；log step 49建立记录并重放该动作的交付流程，log step 77用选定参数生成最终69步交付，记录与重放均掉落。视频1.60–2.76 s可见方块随手腕运动翻转后脱离，手指未将它承接并稳定在目标附近；agent也在log step 87自述缺少把翻转方块接入保持姿势的办法。结合交付与画面，支持调姿后承接、保持的控制失败；证据不足以进一步区分接触布局、轨迹时序或反馈跟踪问题。"
+   }
+  ],
+  "gap_zh": "无逐帧状态日志，无法恢复精确位置误差、朝向余弦及solved_run时间序列，也无接触力与关节跟踪误差，故根因只标控制大类。reward未报告DAG的三个原生指标，native_metrics均留null；log step 77的峰值余弦和log step 87自述约18步仅作辅助证据，不代填评分器指标。节点从未完成由交付总共69步小于必需100步及视频末段失去承托共同确定，无需用绿色锁存指示推断保持时长。"
+ },
+ "summary_zh": "本次单手调姿尝试使黄色方块滚转，视频2.08 s指示变绿，但视频2.40–2.76 s方块脱离手指且未被接住，物体调至目标节点从未完成，终态0/1。log step 77、81确认最终交付只有69个控制步，两次独立重放均因掉落失败，达不到连续100步要求；现象为没保持住，根因支持控制失败。缺少逐帧状态、接触力及跟踪误差，无法细分控制根因或给出精确原生指标。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点（操作中，视频约2.4 s方块脱离）、现象没保持住、根因停在控制大类均有视频与log step 77/81的69步交付重放支持；未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:44:08Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--gd33wrr/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--gd33wrr/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "59faf8cb02ce703a7fc21c888ed75f77f05aaea492aef060c16ac371d5782288",
+   "video_sha256": "c971770ad71151fb3f1a46ce5f3886e0445a9f02b77ca1589be50251acddee9e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--ka2quw7.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--ka2quw7.json
new file mode 100644
index 0000000000000000000000000000000000000000..20efb01240e80e5cf10196464ce942113d69e9d2
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--ka2quw7.json
@@ -0,0 +1,178 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-privileged--ka2quw7",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 164,
+   "sim_steps": 164,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 77,
+   "agent_resets": 15929,
+   "agent_sim_steps": 1598846,
+   "agent_records": 0,
+   "agent_replays": 19,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.18,
+   "peak_processes": 17,
+   "peak_disk_gb": 0.84,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-privileged--ka2quw7/replay.mp4",
+  "duration_s": 6.6,
+  "fps": 25.0,
+  "frames": 165,
+  "sha256": "5dc795f3be01991078d5886b63ff87b4ef95ec590af19413d1c88040ae978cc8"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [
+     [
+      6.56,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 黄色棱角物体已位于单只肌骨手的手指间，目标指示呈红色；视频 0.40–2.40 s 手指收拢并随手腕、前臂转动带动物体调姿，尚未完成连续保持窗口。"
+     },
+     {
+      "t": 2.6,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 2.32–2.80 s，目标指示在视频 2.56 s 仍为红色、视频 2.60 s 首次变绿；随后视频 2.80–6.56 s 物体一直留在手指间，后段朝向趋于稳定，视频 5.52–6.56 s 的逐帧检查未见掉落。绿色仅作瞬时达标线索，不单独作为节点完成证据。"
+     },
+     {
+      "t": 6.56,
+      "log_step": 162,
+      "text_zh": "观察：log step 160 的脚本将最终两阶段肌肉指令生成并保存为交付轨迹；log step 162 对该交付文件两次新进程重放，均报告 success=true、steps=164，终态哈希同为 cfeca5ae587c8da1。inputs/run.json 的评分字段另报告 replay_success=1、deterministic=1、n_actions=164、sim_steps=164；视频 6.56 s 末帧中物体仍被手指保持。"
+     },
+     {
+      "t": 6.56,
+      "log_step": 158,
+      "text_zh": "推断：log step 158 的最终候选记录 first=65、life=164，且其两次重放与 log step 162 的交付重放终态哈希一致。视频共有165帧、25 fps，对应初态加164个控制步；控制步65→视频65/25=2.60 s，已在首次变绿处核对，因此控制步164→视频164/25=6.56 s。结合评分成功，控制步65–164连续100步的保持窗口在视频6.56 s完成，节点从该时刻成立；不是从视频2.60 s便记完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.00–2.80 s 单只肌骨手的拇指与其余手指围住黄色物体，屈指调整接触，并配合手腕、前臂旋转改变物体朝向；视频 2.80–6.56 s 继续在指间保持物体。左右画面是同一只手的不同相机视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "本次交付完成唯一的“物体调至目标”节点：单只肌骨手在指间调姿，并于视频约6.56 s完成连续100个控制步的保持窗口，最终为1/1。评分器判定成功，log step 162 的两次交付重放也均在控制步164成功且终态哈希一致；视频后段的稳定持物表现与成功参考录像相符。无逐帧状态日志，精确朝向余弦与位置误差不能从画面独立量测，完成时刻结合视频变绿事件及交付控制步计数推断；评分字段未报告三个原生指标的数值，均填null。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功运行，唯一节点 object_reoriented 在视频 6.56 s 成立，与评分成功、控制步映射（65→2.60 s 变绿已核对）及交付重放一致；接触标注合理，无失败层。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:43:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--ka2quw7/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--ka2quw7/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "e6539efe686b8fcd66393afd60d63de05367c097907d971ee3ba220ae36e7a85",
+   "video_sha256": "5dc795f3be01991078d5886b63ff87b4ef95ec590af19413d1c88040ae978cc8",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--p8epdyk.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--p8epdyk.json
new file mode 100644
index 0000000000000000000000000000000000000000..80c20036b3fddd601a67ced0898bd3167cbbdd64
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--p8epdyk.json
@@ -0,0 +1,183 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-privileged--p8epdyk",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 251,
+   "sim_steps": 251,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 35,
+   "agent_resets": 25094,
+   "agent_sim_steps": 5692,
+   "agent_records": 0,
+   "agent_replays": 11,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.611,
+   "peak_processes": 15,
+   "peak_disk_gb": 0.245,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-privileged--p8epdyk/replay.mp4",
+  "duration_s": 10.08,
+  "fps": 25.0,
+  "frames": 252,
+  "sha256": "ebeeab0ad91f75aae81a305e64469b0a8e0f241df6b8eb6f19c709a6a06d4755"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [
+     [
+      10.04,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的两个相机视图显示同一只 MyoHand，黄色棱角物体已在手指上，目标指示为红色；初态尚未完成调姿和连续保持。"
+     },
+     {
+      "t": 1.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.20–2.20 s，MyoHand 的拇指与其余手指合拢包住黄色物体，手腕转动，物体在指间改变朝向；后续视频 3.0–6.0 s 物体逐渐稳定在指间，未见脱手。"
+     },
+     {
+      "t": 6.08,
+      "log_step": null,
+      "text_zh": "观察：视频 6.04 s 指示仍红，视频 6.08 s 首次变绿；视频 6.08–10.04 s 两个视图中黄色物体始终留在合拢的手指间，末段朝向基本稳定，已逐帧检查视频 9.00–10.04 s。绿色是瞬时达标的锁存提示，不能单独证明连续保持完成。"
+     },
+     {
+      "t": 10.04,
+      "log_step": 24,
+      "text_zh": "推断：log step 24 对最终交付轨迹的 Session 重放输出 success=True、steps=251、solved_run=100，重新保存后两次独立重放均在控制步251成功且终态哈希相同；log step 28 确认文件仍为251行，log step 31 再次重放成功。视频共252帧、25 fps，符合初帧加每控制步一帧，故控制步k对应视频k/25 s；已用视频 6.08 s 首次变绿（控制步152）核对，控制步152至251恰为100步，完成时刻为视频251/25=10.04 s。此处使用交付轨迹的控制步计数，不使用agent工作时间；视频近似静止的尾段仍属保持过程，不扣除为额外停帧。"
+     },
+     {
+      "t": 10.04,
+      "log_step": 31,
+      "text_zh": "观察：run.json 的评分字段 success=1、replay_success=1、deterministic=1、instance_ok=1、sim_steps=251；log step 31 的评分器两次新进程重放均成功且 state_sha 一致。视频 10.04 s 仍显示黄色物体被手指包住，无直接反证，最终节点判为完成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.20–2.20 s，同一只 MyoHand 的拇指和其余手指合拢包住黄色棱角物体，随手腕转动在指间调整物体朝向；视频 6.08–10.04 s 继续夹持，未见松开或掉落。左右画面是同一只手的两个相机视图。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "本次交付轨迹中，MyoHand 合拢手指并转动手腕，在手内调整黄色物体，视频 6.08 s 首次瞬时达标，结合交付轨迹的成功重放与100步保持计数，唯一节点在视频约 10.04 s 完成，最终为1/1。评分器确认两次重放成功且确定性一致，无失败断点。未提供逐帧状态日志，精确阈值及保持完成时刻由视频与同一交付轨迹的重放计数共同推断；reward 未单列三项原生指标，native_metrics 的数值均留空。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "唯一目标节点在视频 10.04 s 完成：指示在视频 6.08 s 变绿并保持到末帧，控制步152+100=251 的映射与成功重放一致；成功任务无断点，标签无误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:44:59Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--p8epdyk/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--p8epdyk/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1beac7488c8005e579414973b33ac796ea1941aa0b96c759910cafadff77ce3e",
+   "video_sha256": "ebeeab0ad91f75aae81a305e64469b0a8e0f241df6b8eb6f19c709a6a06d4755",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--qp7nfen.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--qp7nfen.json
new file mode 100644
index 0000000000000000000000000000000000000000..c791428dd266a416b43f1148f334635ada97ec78
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-privileged--qp7nfen.json
@@ -0,0 +1,182 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-privileged--qp7nfen",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 146,
+   "sim_steps": 146,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 1121,
+   "agent_resets": 3063,
+   "agent_sim_steps": 89442,
+   "agent_records": 0,
+   "agent_replays": 10,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 122.829,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.263,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_memory_gb": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-privileged--qp7nfen/replay.mp4",
+  "duration_s": 5.88,
+  "fps": 25.0,
+  "frames": 147,
+  "sha256": "fecf72f932c019bb1e9617e019bc5a12e85b721c7493b7ee256b79673b62701d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.84,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00 s，黄色方块已位于单只肌骨灵巧手的手指间，目标示意周围为红色；视频0.00–1.80 s，手腕及手指运动带动方块转动，尚未完成连续保持验收。"
+     },
+     {
+      "t": 1.88,
+      "log_step": null,
+      "text_zh": "观察：视频1.84 s（第46帧）目标示意仍为红色，视频1.88 s（第47帧）变绿；视频1.88–5.84 s中方块始终留在该手的拇指侧指间，朝向总体稳定。绿色仅作首次瞬时达标的线索，不单独证明连续100步完成。"
+     },
+     {
+      "t": 5.84,
+      "log_step": 206,
+      "text_zh": "观察：视频4.80–5.84 s按约0.08 s间隔检查至末帧，黄色方块仍被手指围住，未见脱落；log step 206中生成脚本对最终146行轨迹调用基准Session.replay，返回success=True、steps=146、solved_run=100、hold_steps=100，并保存同一轨迹。"
+     },
+     {
+      "t": 5.84,
+      "log_step": 207,
+      "text_zh": "观察：log step 207对最终保存轨迹执行官方重放工具，两次均返回success=true、steps=146，终态哈希均为07e88d94c7648b00；本次评分reward也记录success=1、replay_success=1、deterministic=1和n_actions=146。"
+     },
+     {
+      "t": 5.84,
+      "log_step": 206,
+      "text_zh": "推断：本视频147帧对应初态加146个交付控制步，25 fps下第146步对应视频146/25=5.84 s；第47步对应视频1.88 s，已用指示首次变绿核对，至第146步恰为100个连续达标步。故把含保持窗口的调姿节点首次完成标在视频5.84 s，终态为真；不把仿真1.0 s保持直接当作视频1.0 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.00–1.88 s，单只灵巧手的拇指侧与其余手指围住黄色方块，配合手腕转动使方块在指间改变朝向；视频1.88–5.84 s继续围持该方块，没有可见松手或换手。左右画面是同一只手的两个视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "value": null,
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better"
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better"
+  },
+  {
+   "name": "solved_run",
+   "value": 100,
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "观察：本次评分成功；视频中单只灵巧手在指间转动黄色方块，随后持续持物至视频5.84 s末帧，完成物体调至目标节点（1/1）。推断：结合最终交付重放的146步、solved_run=100及两次一致的成功结果，将包含保持窗口的完成时刻标为视频5.84 s，首次变绿的 视频1.88 s 尚不是完成时刻。观察：本输入没有逐帧状态日志，精确朝向余弦和位置误差未在最终验收输出中报告，故这两项原生指标留空；连续保持由交付文件的基准重放结果及画面共同支持。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标注与证据一致：视频1.88 s指示变绿，第47–146控制步恰为100步连续达标，节点完成时刻标视频5.84 s合理；log step 206/207的重放结果支持成功，接触标注无误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:45:41Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--qp7nfen/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-privileged--qp7nfen/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "aa268533301cae1e503a4f6d923268293bf0cc8045326fe908fa88d53e4622af",
+   "video_sha256": "fecf72f932c019bb1e9617e019bc5a12e85b721c7493b7ee256b79673b62701d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--5kubwba.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--5kubwba.json
new file mode 100644
index 0000000000000000000000000000000000000000..3b4ed89dee8b76412a65d9effb14c917828450b9
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--5kubwba.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-standard--5kubwba",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 18,
+   "sim_steps": 18,
+   "physics_steps": 90,
+   "steps": 57,
+   "calls": 8,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 57,
+   "elapsed_s": 2776.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.628,
+   "peak_processes": 15,
+   "peak_disk_gb": 0.743,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-standard--5kubwba/replay.mp4",
+  "duration_s": 0.76,
+  "fps": 25.0,
+  "frames": 19,
+  "sha256": "aa3ff40ae8a1b8cd609722d6fcfed4be51682bc96338d73d98971152a916c167"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.00–0.72 s 的全部19帧，单只MyoHand由较伸展姿态逐渐屈指、转腕，黄色物体随手部改变朝向；视频 0.52–0.72 s 仍可见物体位于弯曲手指之间，未见物体脱手落地，也没有参考录像后段那样的持续稳定保持段。"
+     },
+     {
+      "t": 0.72,
+      "log_step": 84,
+      "text_zh": "观察：log step 84 是运行最终gen.py之后的服务状态，episode=3、control_steps=18、done=True、reason=\"failure: object dropped\"；评分器同时报告graded_by_live=1、n_actions=18、sim_steps=18、live_success=0，吻合本视频19帧的单次末局。服务的object dropped表示位置误差超过0.075米，不要求物体落地。"
+     },
+     {
+      "t": 0.72,
+      "log_step": 84,
+      "text_zh": "推断：最终局总共只有18个控制步，故不可能达到连续100个控制步的调姿保持验收，节点全程未完成。18步与视频初帧后18个逐步画面相符；本标注时间均取实际视频帧，不把仿真0.18秒或agent工作时钟当作视频0.72秒。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 0.32–0.72 s，单只MyoHand的拇指与其余弯曲手指从物体两侧围合黄色物体，手腕同时转动；末帧仍见物体在手指间。左右画面是同一只手的两个视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_reoriented",
+   "t": 0.6,
+   "log_step": 84,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.48–0.72 s，MyoHand仍在屈指转腕，黄色物体在手指间随之转动，尚未形成稳定调姿保持；log step 84 的最终现场局由服务以failure: object dropped终止，仅执行18个控制步。推断：物体位置已超出本步允许的0.075米容差，失败发生在调姿操作中；视频 0.60 s 是末段动作的粗略导航锚点。"
+  },
+  "unmet": {
+   "object_reoriented": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.6,
+    "log_step": 84,
+    "evidence_zh": "观察：视频 0.48–0.72 s 中，机器人屈腕并继续合拢手指，黄色物体仍在指间转动；log step 84 的现场服务明确判failure: object dropped。推断：本步操作未把物体位置限制在验收容差内；这是位置失准，画面不足以把该服务用语解释成抓握已经丢失或物体落地。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "观察：log step 80 的本地优化器只按存活控制步数和位置误差优化肌肉序列；log step 82 将所得序列提交服务后在18步终止，log step 83 的最终脚本仍把该序列整段开环提交，段内没有依据图像或本体感知修正动作；log step 84 重置后的最终局再次在18步因object dropped终止。视频 0.00–0.72 s 显示该序列实际驱动手指合拢、手腕转动。推断：所选肌肉控制序列未能在现场维持物体位置，属于控制失败；证据不能进一步区分接触设计、轨迹或反馈控制子类。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志，也未报告最终原生rot_align、norm_obj_err_pos或solved_run数值，故这些指标留null；无法精确还原位置越界过程和各指接触力。目标示意物在画面中有显示偏移，未用其与实物的像素间距计算位置误差。agent在log step 85自述初态穿透、自由落体及“无解”，所据是其自行搭建的本地副本；log step 79、82已记录本地与服务表现不同，不能据此排除所有合法方案或归因任务/环境。"
+ },
+ "summary_zh": "本次评分为最终第3局，MyoHand屈指围合黄色物体并转腕，但仅执行18个控制步即由服务判物体位置超出容差，物体调至目标节点全程未完成（0/1）。断点为调姿操作中的位姿不准，日志支持所选开环肌肉控制序列未能维持位置；视频末帧物体仍在指间，不能把服务的“掉落”直接解释为脱手落地。无逐帧原生状态及指标数值，agent关于初态“无解”的判断未获验证。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "标注与视频、日志一致：最终第3局仅18个控制步即被服务判 object dropped，节点未达成，断点、位姿不准与控制根因均有证据且措辞保守（未把服务用语解读为落地）。未发现需要修改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:07Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--5kubwba/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--5kubwba/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "5868403df4ca16172eec83160ad2ad414d86f65e5cbd08908f6d3469a8fa2e0a",
+   "video_sha256": "aa3ff40ae8a1b8cd609722d6fcfed4be51682bc96338d73d98971152a916c167",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--9q7zktk.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--9q7zktk.json
new file mode 100644
index 0000000000000000000000000000000000000000..0a63cb807e739855ce4edca6f7cf3ac46c03e130
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--9q7zktk.json
@@ -0,0 +1,211 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-standard--9q7zktk",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 79,
+   "sim_steps": 79,
+   "physics_steps": 395,
+   "steps": 648,
+   "calls": 117,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 648,
+   "elapsed_s": 1945.2,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.613,
+   "peak_processes": 9,
+   "peak_disk_gb": 1.178,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-standard--9q7zktk/replay.mp4",
+  "duration_s": 3.2,
+  "fps": 25.0,
+  "frames": 80,
+  "sha256": "60f56a92cca72bb133e58c3b78dc4e65f68869d66dffda356611c2c71bf4bbce"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s，黄色棱角物体由单只灵巧手的伸展手指承托，半透明目标周围为粉红色；视频 0.00–3.16 s 的双视角中，手腕转动，物体未形成参照成功录像后段那样的稳定保持。"
+     },
+     {
+      "t": 2.8,
+      "log_step": null,
+      "text_zh": "观察：视频 2.20–3.16 s，黄色物体相对手指持续滑转并移向指端外侧，与目标的可见间隔增大；末帧未恢复稳定承托，目标指示仍为粉红色。"
+     },
+     {
+      "t": 3.16,
+      "log_step": 113,
+      "text_zh": "推断：评分器记录最后一幕 n_actions=79、sim_steps=79、task_failure=1、live_success=0；结合视频初态未达标及末段外滑，该幕未达到连续100个控制步的保持验收，故本节点全程未完成。agent 在 log step 113 自述最后一幕因 object dropped 于控制步79结束；此自述仅作与画面一致的补充，不作为精确位姿测量。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频 0.00–2.20 s，单只MyoHand的伸展手指从下方承托黄色物体，手腕转动带动物体改变朝向；视频 2.20–3.16 s 物体沿指端外滑。画面未见拇指与其余手指合拢夹紧，不标抓握。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_reoriented",
+   "t": 2.8,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 2.20–3.16 s，机器人仍在转腕调姿，黄色物体沿伸展手指滑转至外侧，未重新稳住，半透明目标指示仍为粉红色。评分器记录最后一幕79个控制步、task_failure=1；agent 在 log step 113 自述终止原因为 object dropped。断点锚定视频约 2.80 s 的外滑过程，精确越过位置容差的时刻未记录。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.00–1.60 s 黄色物体位于手指承托区域；视频 2.20–3.16 s 随机器人转腕，物体相对手指滑转、移到指端外侧，末段没有被重新收回稳住。推断：原有承托未能在调姿过程中保持，构成持物过程中的滑失；无需以物体落地为依据。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 2.8,
+    "log_step": 103,
+    "evidence_zh": "推断：log step 103 的最后一次reset后，agent 脚本按预定序列改变腕部和旋前肌肉命令；log step 104、105、107、108 继续提交短段腕部与屈指肌肉命令。视频 1.60–3.16 s 对应的可见行为是转腕时物体持续移向手指外侧，未通过手指收拢或腕部补偿恢复承托，因此将滑失归于调姿中的接触保持控制不足。现有证据不能进一步区分接触布局、肌肉激活参数或反馈控制子类。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志，评分器字段未报告rot_align、norm_obj_err_pos或solved_run数值，无法量化朝向误差、位置误差和精确越界时刻。部分手指接触被物体遮挡，缺少接触对和接触力，无法确认最终一刻是否仍有局部接触，也无法细分控制根因；log step 103–108 多数调用的输出文本为空，不能把计划打印的状态当成实测反馈。",
+  "unmet": {
+   "object_reoriented": false
+  }
+ },
+ "summary_zh": "最后一幕中，机器人以伸展手指承托黄色物体并转腕调姿，但未完成“物体调至目标”，最终为0/1节点。视频约 2.20–3.16 s 显示物体向指端外侧滑转且未重新稳住，断在调姿操作中，根因归于接触保持控制不足。该幕仅79个控制步，未完成100步保持窗口；缺少逐帧位姿与接触力数据，精确越界时刻及控制子类无法确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "最后一幕79控制步、task_failure=1、成功节点未达成，0/1与断点（操作中、没保持住、根因停在“控制”）均与视频一致：约2.2 s后物体离开手指承托并远离目标。未发现需改动的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:48:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--9q7zktk/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--9q7zktk/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "50b720a0264dd66a720f6abf844b2b726c9c3177e1f7828b00a68ca0f60f8883",
+   "video_sha256": "60f56a92cca72bb133e58c3b78dc4e65f68869d66dffda356611c2c71bf4bbce",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--fupxi2j.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--fupxi2j.json
new file mode 100644
index 0000000000000000000000000000000000000000..b33d44b069b500e56dfb51222ab54678fb70e673
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--fupxi2j.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-standard--fupxi2j",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 77,
+   "sim_steps": 77,
+   "physics_steps": 385,
+   "steps": 587,
+   "calls": 70,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 587,
+   "elapsed_s": 1652.6,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.974,
+   "peak_processes": 12,
+   "peak_disk_gb": 1.492,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-standard--fupxi2j/replay.mp4",
+  "duration_s": 3.12,
+  "fps": 25.0,
+  "frames": 78,
+  "sha256": "91f91f5bdf50aef077e425145c8dc591716baf5056d44c32852fdd49d3896c7e"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.00–2.40 s的双手部视角显示，黄色方块起初靠在单只MyoHand张开的手指与拇指附近，手腕运动时仍在手部承托区；视频2.40–3.08 s方块逐渐转动、滑向手外，末帧与手指明显分离，未出现参考成功录像后段那样持续留在手指间的稳定姿态。"
+     },
+     {
+      "t": 3.08,
+      "log_step": 178,
+      "text_zh": "观察：log step 178明确重置进入最后的episode 11，A2承托控制随后返回done=true、reason=\"failure: object dropped\"、control_steps=77；log step 180、182再次确认同一终态，评分器字段graded_by_live=1、live_success=0、n_actions=77。"
+     },
+     {
+      "t": 3.08,
+      "log_step": 178,
+      "text_zh": "推断：最后评分episode从重置到结束总共只有77个控制步，少于节点要求的连续100个控制步，结合视频全程和末段滑离，节点从未完成保持窗口；这并非仅由零分反推未完成。视频3.12 s的片长不能按真实仿真秒数直接验收保持时长。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "观察：视频0.00–2.40 s，单只MyoHand张开的手指与拇指区域从下方承托黄色方块，手腕调整时方块随接触面运动；视频2.40–3.08 s方块滑离承托区。未见手指合拢包围或夹紧方块。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_reoriented",
+   "t": 2.6,
+   "log_step": 178,
+   "where": "操作中",
+   "evidence_zh": "观察：视频2.40–3.08 s，MyoHand仍在承托调姿时，黄色方块从手指附近滚转、滑向手外，末帧已离开手指承托；log step 178的最后episode 11在A2承托阶段以\"failure: object dropped\"结束，后续log step 180、182确认未恢复。"
+  },
+  "unmet": {
+   "object_reoriented": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.6,
+    "log_step": 178,
+    "evidence_zh": "观察：视频0.00 s黄色方块已由手部承托，视频2.40–2.80 s可见方块沿手指外侧滚转滑移，视频3.08 s右侧视角中方块与伸展的手指之间出现明显间隔；log step 178的服务返回最后episode的结束原因为物体掉落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 178,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 178实际运行的final_run.py在评分episode 11只执行A2承托阶段：每10个控制步按腕屈伸位置及差分速度更新FCR、FCU、PL输入，其余肌肉保持默认输入-1，读取偏摆位置却未用它反馈修正。该次服务本体读数中偏摆由-0.053 rad漂至约-0.18 rad，屈伸位置也未稳定在脚本目标0.15 rad，最终返回物体掉落，与视频2.40–3.08 s承托丧失一致，支持腕手控制未能持续稳住物体；log step 184才写入的屈伸加偏摆改进控制没有在本次评分episode中执行。缺乏接触力与逐帧物体状态，根因保留在控制大类。"
+   }
+  ],
+  "gap_zh": "观察：inputs/没有逐帧状态日志；评分器未报告rot_align、norm_obj_err_pos和solved_run的数值，故三项原生指标均填null。视频与最后episode的服务终止记录足以确认滑离和保持窗口未完成，但无法精确还原每个接触点、力及瞬时朝向余弦，也不足以进一步区分接触布局与反馈控制的贡献。"
+ },
+ "summary_zh": "评分取最后的episode 11：MyoHand已接触并承托黄色方块，但在视频约2.60 s滑移加剧，至视频3.08 s明显离开手指，物体调至目标节点最终为0/1。实际执行的腕屈伸承托控制未能稳住方块，服务在77个控制步后判物体掉落，未完成连续100步保持；结束后写出的改进脚本不属于评分动作。没有逐帧状态日志及原生调姿指标数值，控制根因不细分。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标签与视频和日志一致：最后评分episode 11只有77个控制步即被判物体掉落，节点未达成；断点、接触、现象和停在“控制”的根因均有证据支持。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:44:35Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--fupxi2j/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--fupxi2j/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "2e76c5b3d6ca6c6a2daa642b9c443f2da737055ca347b20193897df4d80c69ac",
+   "video_sha256": "91f91f5bdf50aef077e425145c8dc591716baf5056d44c32852fdd49d3896c7e",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--hyukq3t.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--hyukq3t.json
new file mode 100644
index 0000000000000000000000000000000000000000..b867c4b36476b11a47ced7c37eb14b4e262b75d4
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--hyukq3t.json
@@ -0,0 +1,190 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-standard--hyukq3t",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 149,
+   "sim_steps": 149,
+   "physics_steps": 745,
+   "steps": 286,
+   "calls": 5,
+   "episodes": 2,
+   "resets": 1,
+   "reset_budget": 10,
+   "sim_steps_total": 286,
+   "elapsed_s": 1636.5,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.16,
+   "peak_processes": 25,
+   "peak_disk_gb": 0.381,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-standard--hyukq3t/replay.mp4",
+  "duration_s": 6.0,
+  "fps": 25.0,
+  "frames": 150,
+  "sha256": "f422d74618793045af1ec2c39b260247a8ea80de5e9c8b3e15185eee18559d9c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.96,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的左右画面是同一只肌骨灵巧手的两个视角，黄色方块已在手指间，目标指示为红色；视频 0.00–1.92 s 中手指屈伸并带动物体转动，尚未完成连续保持窗口。"
+     },
+     {
+      "t": 2.0,
+      "log_step": null,
+      "text_zh": "观察：视频 1.92 s 指示仍为红色，视频 2.00 s 已变绿；视频 2.00–5.96 s 的双视角显示黄色物体持续留在手指间，手指与手腕继续小幅调姿。已逐帧检查视频 5.00–5.96 s，末帧物体仍受手指承托和夹持；绿色锁存本身不证明连续保持完成。"
+     },
+     {
+      "t": 5.96,
+      "log_step": 54,
+      "text_zh": "观察：log step 54 是 agent 脚本重置后向服务提交最终动作序列的调用，服务回执为 done=True、reason=success、episode=2、control_steps=149；run.json 的 reward 同时记录 live_success=1、graded_by_live=1、n_actions=149，故评分的是最后这一幕，而非 log step 36 的第一个 episode。"
+     },
+     {
+      "t": 5.96,
+      "log_step": 54,
+      "text_zh": "推断：本次视频为 150 帧、25 fps，对应初始帧和服务的 149 个控制步；末帧视频 149/25=5.96 s 与 log step 54 的成功终止相符。按任务连续100控制步的成功规则，完整验收首次在视频约 5.96 s 达成；仿真100步为1.0秒，在本视频约占4.0秒。参考视频后段同样呈现物体被手指夹持且目标指示变绿，仅用于辨认完成外观。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频 0.00–5.96 s，单只 MyoHand 的手指围住并承托黄色方块，随手指屈伸和手腕转动改变方块朝向及接触位置，物体持续留在手中；左右画面是两个相机视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": null,
+ "summary_zh": "评分采用重置后的第2个episode，灵巧手在手中转动并保持黄色物体，物体调至目标节点于视频约5.96 s完成，最终1/1；log step 54 的服务回执及评分器均确认成功。视频约2.00 s变绿只表示瞬时达标，完成100步保持由成功回执和末段画面共同支持。没有逐帧状态日志，无法独立测量精确朝向余弦、位置误差和连续计数；评分器未报告三个原生指标数值，均留null，未采用本地试验数值替代。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，评分的是重置后的第2个episode（服务回执 done=True、reason=success、control_steps=149）。目标节点在视频约5.96 s完成、final=true、contact_actual（手指手内调整）均与画面和回执一致，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:44:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--hyukq3t/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--hyukq3t/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ebc0622f590dbcfa52e6f96314bc08188c418a1dfdae4f93b4964b40f92a49f2",
+   "video_sha256": "f422d74618793045af1ec2c39b260247a8ea80de5e9c8b3e15185eee18559d9c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--qkxmltc.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--qkxmltc.json
new file mode 100644
index 0000000000000000000000000000000000000000..0e8f3555675918b38dc049305959d0edcc633201
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--qkxmltc.json
@@ -0,0 +1,202 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-standard--qkxmltc",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 850,
+   "sim_steps": 850,
+   "physics_steps": 4250,
+   "steps": 13700,
+   "calls": 27,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 13700,
+   "elapsed_s": 3280.0,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.648,
+   "peak_processes": 18,
+   "peak_disk_gb": 0.575,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-standard--qkxmltc/replay.mp4",
+  "duration_s": 34.04,
+  "fps": 25.0,
+  "frames": 851,
+  "sha256": "f90c143904f7c9fdd414aa469a75d9e3db33c44e1d821a9108ed89d283affbfe"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频0.0 s黄色棱角物体已在单只灵巧手的手指间，旁边目标示意的球形指示为粉红色；视频0.0–6.0 s按0.2 s抽帧可见手腕偏转、手指屈曲，物体随接触转动，属于实际调姿尝试。"
+     },
+     {
+      "t": 8.0,
+      "log_step": 118,
+      "text_zh": "观察：视频6.0–10.0 s的加密帧及视频10.0–33.0 s每秒抽帧显示，黄色物体停在弯曲手指间，目标指示持续粉红色；视频33.0–34.0 s每0.2 s帧（含最后一帧）仍相同，未见掉落。log step 118中agent脚本先后重置并提交ny0、ny1，服务对最后的ny1返回done=False、reason=None；此后日志未再提交或重置。"
+     },
+     {
+      "t": 34.0,
+      "log_step": null,
+      "text_zh": "推断：本幕没有形成目标调姿的连续100控制步保持窗口。视频所见粉红指示始终未转绿，与成功参考录像视频2.4–6.0 s的绿色指示不同；评分器字段live_success=0、task_failure=0、timeout=1与持物停在未达标姿态的画面一致，因此节点全程未完成、终值为假。参考录像仅用于识别达标外观，不作为本幕状态证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.0–6.0 s，单只MyoHand的拇指侧与其余手指接触黄色物体，手腕偏转、手指弯曲使物体在指间转向；视频6.0–34.0 s物体仍由弯曲手指承托、围住。左右画面是同一只手的两个视角。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_reoriented",
+   "t": 8.0,
+   "log_step": 118,
+   "where": "操作中",
+   "evidence_zh": "观察：视频0.0–6.0 s灵巧手实际拨转物体；视频约8.0 s物体已基本停在指间，目标指示仍为粉红色，随后至视频34.0 s未修正为达标状态。log step 118的最后一次重置后提交ny1，服务返回done=False、reason=None；评分器最终live_success=0。本幕失败发生在已启动的调姿操作中。"
+  },
+  "unmet": {
+   "object_reoriented": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 8.0,
+    "log_step": 118,
+    "evidence_zh": "推断：视频0.0–8.0 s可见灵巧手改变物体姿态，但视频8.0–34.0 s物体稳定在指间而达标指示保持粉红，未出现参考录像视频2.4–6.0 s所见的绿色指示；结合log step 118最后一次提交的服务未完成反馈及评分器live_success=0，可判断实际调姿结果未对准验收目标。画面未见物体脱手，不标没保持住。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "无逐帧状态日志，也未报告最后评分幕的原生rot_align、norm_obj_err_pos、solved_run，不能量化朝向误差或保持计数。log step 43给出的拟合目标后来被agent在log step 87自行否定；log step 117–118显示最后选择并提交的是预设稳定姿态候选，但其本地仿真朝向与距离不等于现场测量。缺少可信目标朝向和实际物体状态对照，无法区分目标估计错误与肌肉控制能力不足，故根因留空；不采纳agent在log step 124声称目标超出可达集合的结论。视频851帧对应本幕850控制步，100 Hz控制、25 fps录像约为四倍慢放；未将agent会话时间或本地测试计数当作视频时间。"
+ },
+ "summary_zh": "评分幕是log step 118最后一次重置后的ny1尝试：灵巧手实际接触并拨转黄色物体，随后持物停稳，但唯一调姿节点未完成，终态0/1。断点在调姿操作中，视频约8.0 s后至视频34.0 s仍停在未达标姿态，现象为位姿不准，未见掉落。缺少现场朝向、位置误差与连续保持计数，日志不足以确定感知或控制根因；先前会话测试及成功参考均不计入本幕进度。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未完成、终值0、断点object_reoriented/操作中、现象位姿不准均与视频和日志一致；根因留空并附gap_zh，可接受。视频末帧物体仍在手指间、目标指示仍粉红，未见掉落。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:47:48Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--qkxmltc/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--qkxmltc/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "b4ac356ccff3e69824022e8b819fe72f116b3faa1dccfb4e5dd1899a8f1f216d",
+   "video_sha256": "f90c143904f7c9fdd414aa469a75d9e3db33c44e1d821a9108ed89d283affbfe",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--vgseuog.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--vgseuog.json
new file mode 100644
index 0000000000000000000000000000000000000000..2b7e5a8765d7e2d3d9fb3be66d484cdf5bf918f0
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--vgseuog.json
@@ -0,0 +1,194 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-standard--vgseuog",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "physics_steps": 0,
+   "steps": 0,
+   "calls": 1,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 0,
+   "elapsed_s": 3703.6,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.727,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.414,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-standard--vgseuog/replay.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "7c31bff223eceef48c167de6fd09e0b097ac90a00d5a981fa99474a6be1a9aa6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 3,
+      "text_zh": "观察：本次视频仅有视频 0.00 s 的一帧，两个手部视角中黄色物体仍在灵巧手指间，目标周围为粉红色指示区，画面未包含任何调姿运动；log step 3 的机器人服务状态为 control_steps=0、episode=1。"
+     },
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "推断：评分器记录 n_actions=0、sim_steps=0、physics_steps=0、resets=0，且 live_matches_replay=1；结合视频 0.00 s 的初始画面，评分局未执行调姿，也未推进连续100个控制步的保持窗口，故节点始终未完成。该判断结合零动作记录与验收窗口，并非仅由 success=0 推出。"
+     },
+     {
+      "t": null,
+      "log_step": 64,
+      "text_zh": "观察：log step 43、47、51、55、63 中 agent 的脚本在自建 MuJoCo 模型中优化肌肉或姿态控制；log step 64 的 agent 最后自述为已启动本地姿态PD参数优化并等待结束。上述会话测试没有提交到机器人服务，不能计入本次视频的节点进度或实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "value": null,
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better"
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better"
+  },
+  {
+   "name": "solved_run",
+   "value": null,
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_reoriented",
+   "t": 0.0,
+   "log_step": 3,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0.00 s 是评分录像的唯一一帧，灵巧手与物体停在初始构型；log step 3 的机器人服务返回 control_steps=0，评分器最终仍记录 n_actions=0、sim_steps=0、episodes=1、resets=0、timeout=1、infra_error=0。评分局没有开始物体调姿，断点定位在该步启动之前。"
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 3,
+    "evidence_zh": "观察：本次视频 0.00 s 的唯一画面只显示初态；机器人服务在 log step 3 返回 control_steps=0，评分器最终 n_actions=0、physics_steps=0。agent 的本地仿真测试未驱动评分局中的灵巧手，调姿步骤根本未启动。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "观察：本次输入无状态日志，评分器未报告 rot_align、norm_obj_err_pos、solved_run 的原生数值，三个指标均填 null。log step 50、54、58、59、62 的不稳定或失准输出属于 agent 自建模型的会话测试；log step 64 只记录等待最后一次本地优化，未提供该优化的最终结果或未向服务提交动作的明确原因。推断：这些记录不足以将评分局未启动归因于特定的规划、感知、控制或交付错误，因此根因留空；timeout 仅记录结束方式，不作根因。",
+  "unmet": {
+   "object_reoriented": false
+  }
+ },
+ "summary_zh": "评分局仅保留视频 0.00 s 的初始帧，物体调至目标节点未完成，终态进度为0/1。机器人服务与评分器均记录零控制步、零动作，故断点为物体调至目标的首步前，现象为没尝试。agent 在会话中进行了多轮自建模型控制优化，但未提交到评分局；最后优化结果和未提交的具体原因未见记录，根因留空。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "评分局仅有视频 0.00 s 一帧，n_actions=0、control_steps=0；节点 object_reoriented 未达成、断点首步前/没尝试、根因留空并有 gap_zh，均符合 03 规则（agent 自身超时不属外因，会话本地测试不计入评分局）。未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:49:23Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--vgseuog/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--vgseuog/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1dabc447f496d24986050e954c142fa0fdbb024275b2c0cf1b7390d05a41b112",
+   "video_sha256": "7c31bff223eceef48c167de6fd09e0b097ac90a00d5a981fa99474a6be1a9aa6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--yflsmbm.json b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--yflsmbm.json
new file mode 100644
index 0000000000000000000000000000000000000000..4323ccb197caf8c2690f4f9fbadf2a0d703b90ef
--- /dev/null
+++ b/phase2/myosuite/obj_reorient/trajectories/myosuite-obj-reorient-i00-standard--yflsmbm.json
@@ -0,0 +1,211 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-obj-reorient-i00-standard--yflsmbm",
+ "taskKey": "myosuite/obj_reorient",
+ "dag_sha256": "5fa14ac531d86a618258cac5dcf345b855048938d31fd6daf66194ddd0c11023",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 35,
+   "sim_steps": 35,
+   "physics_steps": 175,
+   "steps": 242,
+   "calls": 29,
+   "episodes": 8,
+   "resets": 7,
+   "reset_budget": 10,
+   "sim_steps_total": 242,
+   "elapsed_s": 3427.3,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.469,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.385,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-obj-reorient-i00-standard--yflsmbm/replay.mp4",
+  "duration_s": 1.44,
+  "fps": 25.0,
+  "frames": 36,
+  "sha256": "a54df20e6b25a162fce50bd133f31eb2b5826f3e3f153c311872b5668c616bbf"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "object_reoriented": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 509,
+      "text_zh": "观察：视频0.00 s初帧中，黄色方块位于单只MyoHand的手指旁，目标指示为粉红色；视频0.00–1.40 s全部36帧中，机器人屈伸手指、后仰手腕并拨转方块，末帧方块仍在手指边，未见脱手下落，目标指示始终未变绿。log step509记录重置到episode=8、control_steps=0；此前各局不是本次评分轨迹。"
+     },
+     {
+      "t": 1.4,
+      "log_step": 540,
+      "text_zh": "观察：log step540中，agent只向第8局提交true_winning_actions的前35个控制步，服务返回control_steps=35、done=False、reason=None；log step541和546再次确认仍为35步。视频0.80–1.40 s中黄色方块继续在手指边转动，尚未呈现目标朝向的稳定保持；已对照参考视频后段的持续持物调姿结果，参考不作为本次进度证据。"
+     },
+     {
+      "t": 1.4,
+      "log_step": 546,
+      "text_zh": "推断：log step8所载服务任务要求连续100个控制步达标，而本次第8局总共只运行35个控制步，连完整保持窗口都不足，因此节点全程未完成、终值为false；这也与评分器success=0、live_success=0一致。此判断依据本次控制步数和所见轨迹，不将视频1.44 s当作仿真保持1.0 s，也不采用agent本地仿真的成功计数。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "手内调整",
+      "note_zh": "观察：视频0.20–1.40 s中，单只MyoHand的手指屈曲，贴着黄色方块承托、拨转，并随转腕改变方块姿态；两个画面是同一只手的不同视角，不是双手协作。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rot_align",
+   "unit": "1",
+   "threshold": 0.95,
+   "direction": "higher_better",
+   "value": null
+  },
+  {
+   "name": "norm_obj_err_pos",
+   "unit": "m",
+   "threshold": 0.075,
+   "direction": "lower_better",
+   "value": null
+  },
+  {
+   "name": "solved_run",
+   "unit": "control_steps",
+   "threshold": 100,
+   "direction": "higher_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "object_reoriented",
+   "t": 0.8,
+   "log_step": 540,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约0.80 s时机器人已后仰手腕、屈指拨转黄色方块，但方块仍呈偏离目标示意的倾斜姿态，至视频1.40 s末帧也没有进入稳定的目标保持；此处是持续调姿偏差的粗略导航锚点，没有可定位的突发掉落事件。log step540与546表明评分局止于35个控制步，未完成连续100步验收。"
+  },
+  "unmet": {
+   "object_reoriented": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.8,
+    "log_step": null,
+    "evidence_zh": "观察：视频0.40–1.40 s中，机器人手腕后仰、手指收拢，黄色方块随之转动；末段方块的倾斜方向仍与目标示意不同，粉红指示始终保留，没有出现对准目标后的持续保持过程。黄色方块仍接触手指，画面没有显示本次物体掉落。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 495,
+    "t": null,
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent使用预计算的开环肌肉激活序列，但尚未得到能在服务中持续承托并调姿的控制。log step490、492、495显示同一true_winning_actions候选在第7局以hold方式执行，服务到control_steps=41即报object dropped；log step538的本地试验将候选在控制步35后固定，也仅存活54个控制步。随后log step540在本次第8局仅执行该候选的前35步，视频0.40–1.40 s显示方块尚未调到目标；这些试验支持一般控制能力不足，不能确定更细的控制子类。前一局掉落仅用于根因分析。"
+   }
+  ],
+  "gap_zh": "本次没有状态日志，评分器未报告rot_align、norm_obj_err_pos或solved_run，故三项原生指标均为null，不能量化末态朝向误差、位置误差或瞬时达标时刻。视频可确认实际接触及未完成保持窗口；本地模拟与先前服务试验结果不同，现有证据不足以区分接触设计、轨迹、反馈或参数问题，也不能将agent在log step547自述的腕关节限位机制视作本次已证实根因。"
+ },
+ "summary_zh": "评分的第8局中，单只MyoHand实际承托、拨转了黄色方块，但只执行35个控制步，尚未完成目标朝向连续100步的保持，唯一节点始终未达成（0/1）。断点为物体调姿操作中，现象是位姿不准；同一候选在先前服务试验中的失败支持一般控制问题，本次视频没有显示物体掉落。缺少逐帧状态和原生误差、保持计数，无法细分控制根因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频、日志一致：评分局仅35个控制步（log step 540/546），视频全程方块仍在手指旁、目标指示未变绿，节点未达成、contact_actual为手内调整合理；断点、位姿不准和控制根因均有日志（step 495 第7局41步掉落）与视频支持，未见需要改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:47:32Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--yflsmbm/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-obj-reorient-i00-standard--yflsmbm/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "4591e87f08f721eef6665b376a93abed26ce2e43ba0d49dda13398712ed77176",
+   "video_sha256": "a54df20e6b25a162fce50bd133f31eb2b5826f3e3f153c311872b5668c616bbf",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--5swn68y.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--5swn68y.json
new file mode 100644
index 0000000000000000000000000000000000000000..6cefd6ab3da9454d274f58983a6e50121d1f8fe9
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--5swn68y.json
@@ -0,0 +1,167 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-privileged--5swn68y",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 343,
+   "sim_steps": 343,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 353,
+   "agent_resets": 2655,
+   "agent_sim_steps": 522795,
+   "agent_records": 0,
+   "agent_replays": 5,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 11.568,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.846,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 2,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_oom": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-privileged--5swn68y/replay.mp4",
+  "duration_s": 13.76,
+  "fps": 25.0,
+  "frames": 344,
+  "sha256": "03ff6c28028102b44ea091d6d5513b962b1f1b8bef867ac603cd726f2dfb6f42"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 13.72,
+      "log_step": null,
+      "text_zh": "观察：视频全程按约0.5秒取帧检查，并对视频12.72–13.72 s逐帧检查；画面主要是跑道，模型只在末段画面上缘露出局部，无法看清全身、足地接触或终点到达。"
+     },
+     {
+      "t": null,
+      "log_step": 208,
+      "text_zh": "观察：log step 208 对最终交付文件进行两次新进程回放，两次均为success=false、steps=343、replayed_rows=343、reason=null，终态哈希均为2f085e9714ef3e64；评分器字段replay_success=0、deterministic=1、instance_ok=1与此一致。"
+     },
+     {
+      "t": null,
+      "log_step": 208,
+      "text_zh": "推断：本任务在原生solved首次成立时即成功结束，因此最终文件的两次完整回放均未成功，表明该event节点从未达到；判定依据是该文件的原生回放结果，而非总奖励为零。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频0–13.72 s未显示可辨认的脚掌或假肢足端与跑道的接触区域；末段上缘仅见局部模型，无法确认实际支撑方式。log step 208确认已有动作序列交付，故不标为未尝试或不适用。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 208,
+   "where": "操作中",
+   "evidence_zh": "推断：log step 197的生成器为跑步搜索肌肉动作并保存已执行前缀，log step 207确认最终文件有343行，log step 208两次完整回放均未触发成功；因此失败位于正在尝试的“跑到终点”。视频全程缺少完整身体视图，不能定位某次失衡或其他不可恢复偏差的时刻，t留空；reason=null也不能证明原生跌倒分支已触发。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "观察：inputs/有视频和完整会话日志，但没有状态日志或成功参考视频；视频0–13.72 s主要显示跑道，末段逐帧及上缘放大仍看不清全身姿态和足地接触，不能可靠区分平衡、步态或其他物理现象，故现象留空。log step 208的最终文件回放只报告未成功，reason=null，没有具体失效分支或前向距离；agent在log step 208自述其他较长前缀曾摔倒、约前进1.4米，不能替代本343行交付轨迹的观测。会话包含多轮控制试验及恢复快照，缺少将某一控制缺陷与本次评分轨迹联系起来的充分证据，根因留空，原生距离指标填null。",
+  "unmet": {
+   "track_finished": false
+  }
+ },
+ "summary_zh": "已交付343个肌肉控制动作的跑步尝试，两次完整回放均未触发原生成功事件，“跑到终点”最终为0/1。断点属于“跑到终点”的操作中，但视频主要显示跑道且模型几乎在画外，不能确认具体失效时刻、足地接触或失败现象。无状态日志和原生距离数值，会话其他前缀的摔倒与距离自述不作为本次评分轨迹的事实，根因暂留空。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注保守但有据：视频几乎只显示跑道，模型基本在画外，无法判定姿态/足地接触；log step 208 两次回放 success=false、343 行，track_finished 为 false，断点与空现象/根因配 gap_zh 合理，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:49:15Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--5swn68y/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--5swn68y/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "abcf5597698486012c5ecac4f98a0e149973e1d6808b70b8cb2aefe5a30d96da",
+   "video_sha256": "03ff6c28028102b44ea091d6d5513b962b1f1b8bef867ac603cd726f2dfb6f42",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--b8rdc4u.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--b8rdc4u.json
new file mode 100644
index 0000000000000000000000000000000000000000..ee752809de47c48387624288435d013b3338f885
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--b8rdc4u.json
@@ -0,0 +1,170 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-privileged--b8rdc4u",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 281,
+   "sim_steps": 281,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 374,
+   "agent_resets": 908,
+   "agent_sim_steps": 33956,
+   "agent_records": 31,
+   "agent_replays": 9,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 11.729,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.266,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 3,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_oom": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-privileged--b8rdc4u/replay.mp4",
+  "duration_s": 11.28,
+  "fps": 25.0,
+  "frames": 282,
+  "sha256": "1abf62b700f5cfb716bd458e2359eca0393a75fc8ad448659f7772a70b64d303"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 11.24,
+      "log_step": 400,
+      "text_zh": "观察：视频 0.0–11.24 s 的抽帧及视频 10.28–11.24 s 的逐帧检查主要显示空跑道；上缘仅露出极小的模型局部，无法辨认完整跑姿、足部支撑或终点关系。log step 400 对交付的 trajectory.npz 两次原生回放均报告 success=false、steps=281、state_sha=2993a377aad628a4，且 instance_ok=true、deterministic=true；评分器字段 replay_success=0、task_failure=1。"
+     },
+     {
+      "t": null,
+      "log_step": 282,
+      "text_zh": "推断：log step 282 明确加载交付文件的全部281条动作，回放末尾服务 status 为 solved=0、solved_run=0、done=true，终态哈希与 log step 400 相同。DAG采用首次 solved 即终止的事件验收，因此“跑到终点”从未成立，区间为空；此判断来自交付文件的原生验收记录，并非仅由零分推定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 0.0–11.24 s 中跑者主体及双足基本在画面上缘之外，末段仅有很小的模型局部；没有状态日志或可见的足端—跑道接触，不能确认左脚及右侧假肢足端的实际支撑方式。交付文件包含行进尝试，故保留未知接触而不省略。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 400,
+   "where": "操作中",
+   "evidence_zh": "推断：log step 282 明确回放交付的281条行进动作，log step 400 的两次原生回放均在281步以 failure: fallen, off the track or out of time 结束，未触发 solved；唯一正在执行的语义步骤为“跑到终点”。视频 0.0–11.24 s 无法显示完整身体，不能定位失效的视频时刻，也不能仅凭这个合并失败字符串确定具体失败分支。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 310,
+    "evidence_zh": "推断：作为控制根因的会话证据，log step 282 和 log step 310 的诊断脚本均从初态执行交付文件，终态哈希2993a377aad628a4与 log step 400 的原生回放一致。log step 282 记录模型根部高度由控制步120的0.97 m降至控制步280的0.23 m；log step 310 记录控制步265–280的头部高度由0.643 m连续降至0.348 m。log step 331–332 显示生成动作的脚本使用参考步态跟踪和躯干姿态反馈，却未维持身体高度与支撑；据此归为行进控制未能稳定身体。现有证据不足以进一步确定反馈律、参考轨迹、肌肉力分配或某个参数哪一项是直接原因。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "视频确实存在，但已检查的视频 0.0–11.24 s 的全程半秒抽帧、上缘放大图及视频 10.28–11.24 s 的逐帧画面中，跑者主体和足部大多在画面上缘之外；缺少状态日志及成功参照视频。交付回放诊断提示身体高度下降，但不能以会话诊断替代可见的评分回合失败现象，故现象层留空、实际接触填未知、断点视频时刻留空。合并失败原因没有逐分支服务判定，未将其直接标成跌倒或出界；也未将仿真控制步换算为视频时间。评分器未报告 forward_distance_m；log step 401 的脚本自算距离不作为原生评分指标。"
+ },
+ "summary_zh": "交付的281条动作经两次原生回放均未触发 solved，唯一节点“跑到终点”未达成，最终为0/1。断点位于该行进步骤操作中；同一交付文件的会话诊断显示身体高度持续下降，支持宽泛的控制根因。视频中的跑者主体与足部基本在画面外，无法直接核实具体失败现象、实际接触或失效视频时刻，故相应留空或标未知。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "唯一节点“跑到终点”未达成、final=0 与 log step 400 的两次原生回放（success=false、281步、哈希一致）和评分字段相符；视频只有空跑道，现象层留空并附 gap_zh、接触填未知、断点时间留空均有依据，控制根因有交付文件回放诊断支撑，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:48:47Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--b8rdc4u/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--b8rdc4u/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "631ba4ce92fc1b77e953aca511729a556c9ad693ae1b73cd9e9b6b069374e3ff",
+   "video_sha256": "1abf62b700f5cfb716bd458e2359eca0393a75fc8ad448659f7772a70b64d303",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--h2n7yzw.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--h2n7yzw.json
new file mode 100644
index 0000000000000000000000000000000000000000..f945032796cabb6ab09433623d7a63420cdd1470
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--h2n7yzw.json
@@ -0,0 +1,184 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-privileged--h2n7yzw",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 671,
+   "sim_steps": 671,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 289,
+   "agent_resets": 291,
+   "agent_sim_steps": 19703,
+   "agent_records": 52,
+   "agent_replays": 8,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 3.242,
+   "peak_processes": 23,
+   "peak_disk_gb": 0.247,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-privileged--h2n7yzw/replay.mp4",
+  "duration_s": 26.88,
+  "fps": 25.0,
+  "frames": 672,
+  "sha256": "b940852194b886511adc02cbe0172b9c63692c7886a7c0742e2051fa30c07154"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 12.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0–26.84 s 的全程抽帧及上缘放大中，跑者从画面上缘逐渐露出，视频 12.5–24.5 s 可见腿部形体交替变化并前移；视频 25.56–26.84 s 形体转斜、展开，未见到达终点的画面。"
+     },
+     {
+      "t": 26.84,
+      "log_step": 69,
+      "text_zh": "观察：log step 64 将 671 步动作数组保存为最终交付文件；log step 69 对该文件进行两次原生回放，均在 671 步返回 success=false、相同 state_sha 及 failure: fallen, off the track or out of time。评分器 reward 的 replay_success=0、task_failure=1、deterministic=1，与视频末段一致。"
+     },
+     {
+      "t": null,
+      "log_step": 69,
+      "text_zh": "推断：此任务的原生 solved 首次成立即成功结束回合，而交付文件的两次完整回放均以失败结束，故“跑到终点”在本回合未达到；此判断结合原生回放结果，不从零分单独反推。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 6–26.84 s 上缘放大可见跑者移动及腿部形体变化，但身体仅占少量像素，脚底与跑道的接触面不能分辨；无状态日志，无法核实左脚、右假肢足端分别何时接地及其实际支撑方式。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": 25.6,
+   "log_step": 69,
+   "where": "操作中",
+   "evidence_zh": "推断：视频 12.5–24.5 s 跑者仍在行进，视频约 25.6–26.84 s 身体形体由狭长转为明显转斜、展开，之后视频结束；结合 log step 69 交付文件两次回放失败，以及 agent 在 log step 70 自述该交付轨迹最终跌倒，判定未修复的失败发生在跑到终点的行进过程中。视频约 25.6 s 仅为末段姿态变化的导航锚点，远景不足以定位精确失去支撑的一帧。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 25.6,
+    "log_step": 69,
+    "evidence_zh": "推断：视频 25.44–26.84 s 放大连续帧中，跑者的深色狭长形体转斜并缩成展开的肢体投影；结合 log step 64 保存的最终轨迹、log step 69 对同一交付文件的两次失败回放，以及 agent 在 log step 70 自述“then falls”，支持末段失衡而未完成跑道。画面分辨率不足以直接确认哪条腿先失去支撑，未将该细节作事实。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 63,
+    "source_zh": "日志 / 视频 / agent 自述",
+    "evidence_zh": "推断：log step 56 的脚本通过支撑髋的骨盆俯仰调节和摆动髋目标修正来维持跑步；log step 57–60、63 的多个控制配置仍以失败结束。log step 63 的 671 步配置随后被 log step 64 保存并由 log step 69 两次回放验证失败，视频 25.6–26.84 s 也呈现末段姿态塌乱；agent 在 log step 70 将其归于假肢摆动与身体支撑未协调。证据支持现有肌肉与步态控制未维持动态平衡，但不足以证明某一增益、接触布局或具体反馈环节单独致因，故停在“控制”。"
+   }
+  ],
+  "gap_zh": "观察：inputs/ 未提供状态日志或成功参考视频；现有视频为 320×240 的远景，跑者早段多在画面上缘以外，后段仍只有少量像素，不能可靠辨认脚底接触、失衡方向或精确跌倒触发帧。原生回放的失败原因是多个条件共用的文本，没有独立失败分支字段；“平衡”由末段形体变化与已交付轨迹的日志交叉推断。评分器未报告 forward_distance_m，故指标值为 null；agent 在 log step 70 自述的仿真约 3.4 s 及约 2.6 m 不当作视频时间或原生测量值，未据此换算视频时刻。"
+ },
+ "summary_zh": "交付的 671 步轨迹已开始沿跑道行进，但“跑到终点”未达到，最终节点为 0/1；log step 69 的两次原生回放均确定失败且终态一致。视频约 25.6–26.84 s 的形体变化与交付日志共同支持末段失衡，根因保留为控制未能维持动态平衡。远景分辨率限制实际接触及具体控制失效环节的判断，无状态日志且评分器未报告前向距离。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "track_finished 未达成（final 0）、断点在该步骤的行进过程中，与 log step 69 的两次回放失败（671 步）及视频末段（约 24.4–26.84 s 跑者形体转斜、移出画面上缘）一致；平衡现象与控制根因都标为推断且措辞克制，视频时间未由 agent 时间换算。未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:46:51Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--h2n7yzw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--h2n7yzw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f77046815f530613e7297aff3d82039b4799c6ba778d93df3d08182be3213e7b",
+   "video_sha256": "b940852194b886511adc02cbe0172b9c63692c7886a7c0742e2051fa30c07154",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--hemxnsc.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--hemxnsc.json
new file mode 100644
index 0000000000000000000000000000000000000000..5c8d03ff9e5074116d6e4590257c8588a7f5355a
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--hemxnsc.json
@@ -0,0 +1,177 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-privileged--hemxnsc",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 237,
+   "sim_steps": 237,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 2256,
+   "agent_resets": 2263,
+   "agent_sim_steps": 14882,
+   "agent_records": 3,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 126.218,
+   "peak_processes": 20,
+   "peak_disk_gb": 0.244,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_memory_gb": 1,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-privileged--hemxnsc/replay.mp4",
+  "duration_s": 9.52,
+  "fps": 25.0,
+  "frames": 238,
+  "sha256": "fc3f27196c6e475f34c7d87abc6406d9711affdf0a5c84edb7fa5e257f190e82"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 80,
+      "text_zh": "观察：agent 的生成脚本从冻结初态运行调参后的肌肉控制器，录制、回放并保存交付轨迹；该次运行和回放均报告 success=false、done=true、steps=237，结束原因仅为合并提示“failure: fallen, off the track or out of time”。"
+     },
+     {
+      "t": null,
+      "log_step": 82,
+      "text_zh": "推断：针对交付文件的检查器回放报告 n_actions=237、replayed_rows=237、success=false、instance_ok=true、deterministic=true，与 run.json 的 replay_success=0、task_failure=1 一致；本任务在首次原生 solved 时即成功结束，因此“跑到终点”事件未达到，并非仅由 reward=0 推断。"
+     },
+     {
+      "t": 9.48,
+      "log_step": null,
+      "text_zh": "观察：已查看视频全程每约0.5秒的帧，并加密检查首段及视频8.40–9.48 s逐帧；画面主要为固定远景跑道，末段上缘仅露出极少量粉白色肢体样片段，看不到完整跑者、足地接触或可核对的29米位移。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频0.00–9.48 s的已检帧不能辨认左脚、右侧假肢足端与跑道的实际接触；无状态日志，不能把DAG的典型支撑接触当成本次事实。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 82,
+   "where": "操作中",
+   "evidence_zh": "推断：log step 80 显示 agent 交付的是实际执行肌肉步态控制器直至服务判定 done 的237步轨迹，log step 82 对该文件的确定性回放仍未触发成功，故断点为唯一行进步骤“跑到终点”的操作中；视频无法辨认完整身体运动，不能定位失效的具体视频时刻。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 80,
+    "source_zh": "日志",
+    "evidence_zh": "推断：会话诊断 log step 44 记录早期控制器测试中头部相对足端高度持续降低至0.181米，log step 46 记录另一诊断中假肢膝角增大且重心高度下降；这些是会话测试，不能当作交付视频的跌倒现象。随后同一控制器族经过相位、反馈增益及姿态参数搜索仍反复提前失败（log step 62、68、74、79），最终录制的控制器也在237步触发失败，且交付文件重放一致（log step 80、82）；这些证据支持未形成可持续行进的肌肉控制这一宽泛归因，不能区分接触设计、反馈控制或某个参数的具体责任。"
+   }
+  ],
+  "gap_zh": "观察：inputs/中有运行记录和9.52秒视频，但没有状态日志或成功参考视频；已检全程粗帧、顶部放大图和视频8.40–9.48 s逐帧，完整跑者及足地接触始终不可辨。交付回放只给出跌倒、出界或时间结束的合并原因，没有独立失败分支、最终前向距离或身体高度；因此现象层留空，contact_actual为未知，原生距离值为null，断点视频时刻为null。agent 在 log step 87 自述交付轨迹约仿真1.2 s跌倒及假肢膝屈曲机制，缺少交付状态数据和可核对的可见事件，未将此自述转为视频时间或确定现象；会话诊断仅用于宽泛控制根因。"
+ },
+ "summary_zh": "观察：交付轨迹包含237步肌肉控制命令，检查器确定性回放报告失败，唯一节点“跑到终点”未完成（0/1；log step 80、82）。推断：断点为该行进步骤的操作中，会话中多次姿态塌落诊断及调参后仍提前失败支持宽泛的控制问题，但不足以确定交付轨迹的具体失效机制。观察：视频主要显示跑道，完整跑者及足地接触不可辨，故现象留空、实际接触未知，未报告的前向距离保留null。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "唯一节点 track_finished 未达成、断点、空现象层（附 gap_zh）与宽泛控制根因均有 log step 80/82 的检查器回放结果支撑；视频只显示固定远景跑道，没有可辨认的跑者，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:46:50Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--hemxnsc/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--hemxnsc/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6c32547db1a749ffef339a05849dcf17098ad077ae1d6b17d8411831cea61273",
+   "video_sha256": "fc3f27196c6e475f34c7d87abc6406d9711affdf0a5c84edb7fa5e257f190e82",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--ifembbr.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--ifembbr.json
new file mode 100644
index 0000000000000000000000000000000000000000..58ff26f6e22a4bf6c20e28c9ecf21200956d856d
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--ifembbr.json
@@ -0,0 +1,181 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-privileged--ifembbr",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 767,
+   "sim_steps": 767,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 101,
+   "agent_resets": 4028,
+   "agent_sim_steps": 25042,
+   "agent_records": 23,
+   "agent_replays": 7,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-privileged--ifembbr/replay.mp4",
+  "duration_s": 30.72,
+  "fps": 25.0,
+  "frames": 768,
+  "sha256": "88f832a8e00d06087521aac826a1ec601f5df0a18f3e67c2bd13f270d7ada3a5"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0–30.68 s 的全景和上沿放大帧中，跑者从画面上沿逐渐进入，仅在跑道远端作短段前移和迈腿；视频 29.60–30.68 s 已逐帧检查，身体仍很小，足底与躯干姿态不能可靠辨认，画面没有可读距离标尺。"
+     },
+     {
+      "t": null,
+      "log_step": 128,
+      "text_zh": "观察：log step 121 的选择脚本经 log step 124 加入 phase3 候选、log step 126 再次执行；log step 128 输出所选 phase3 候选为 767 步、脚本计算 distance=2.2875533280346048，随后对保存动作回放得到 success=False、done=False、solved=0。评分器 reward 同样记录 n_actions=767、replay_success=0、deterministic=1、instance_ok=1。"
+     },
+     {
+      "t": null,
+      "log_step": 128,
+      "text_zh": "推断：本节点是原生 solved 首次成立即结束的事件；最终交付的 767 步没有触发该事件，且视频只支持短段行进，因此没有达到“跑到终点”。不把脚本计算的距离当作评分器原生指标，也不把会话中其他候选的跌倒记入此交付。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 10–30.68 s 的远景上沿可见跑者迈腿与前移，但像素不足，部分身体被上沿裁切，无法确认哪只脚或假肢足端在何时接触跑道；没有状态接触日志，故实际支撑接触标为未知。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 128,
+   "where": "操作中",
+   "evidence_zh": "推断：失败发生在“跑到终点”的行进尝试中。视频 10–30.68 s 可见跑者在远端迈腿前移，log step 121/126/128 表明 agent 选择并保存了 767 步的短动作前缀，原生状态仍为 solved=0、done=0；评分器确认该交付回放失败且一致。未见可可靠定位的不可逆跌倒或出界时刻，故断点时间留空，不用其他候选的失衡时刻代替。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "步态",
+    "t": 30.68,
+    "log_step": 128,
+    "evidence_zh": "推断：本次行进距离不足，交付在跑完前结束。视频 8–30.68 s 的上沿放大序列仅显示跑者从远端逐渐进入画面、迈腿和有限前移；log step 128 对最终 767 步候选计算的 distance 约为 2.29 m，回放状态 solved=0、done=0，与评分器 replay_success=0 相符。此处标短程行进未完成，不据会话延长测试的跌倒认定本视频已摔倒。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent 的肌肉控制器未能生成持续跑完整条跑道的动作。log step 116 的后段步态搜索会在根位置高度低于 0.45 或横向超出自定界限时截断，log step 124–125 的 phase3 细化仍只得到根位置高度约 0.446 的短候选；log step 121 的最终选择脚本又按根位置高度 0.6 等条件截取前缀，log step 126/128 保存并回放的最终结果仅 767 步、solved=0。agent 在 log step 127 也明确承认没有候选到达终点。控制失败的会话证据解释了为何交付仅含短程动作；没有证据把问题进一步归为某个增益、逆解或反馈子类。"
+   }
+  ],
+  "gap_zh": "观察：inputs/ 内有完整视频和运行日志，但没有状态日志或成功参考视频。远景中跑者只占画面上沿少量像素，无法确认足底接触、失衡起点或原生跌倒分支；因此接触标未知、断点时间留空，不把其他测试中的跌倒当作本次现象。视频没有可读距离标尺；约 2.29 m 仅来自 log step 128 的 agent 选择脚本，reward 未报告 forward_distance_m，故该指标值留空。",
+  "unmet": {
+   "track_finished": false
+  }
+ },
+ "summary_zh": "观察：最终交付为 767 步的短行进前缀，评分器回放一致但未成功；“跑到终点”未达到，最终节点为 0/1。推断：失败在行进操作中，现象为步态未完成所需距离；会话中的后段步态搜索及最终截取记录支持控制层面的原因。观察：远景无法可靠辨认支撑接触或失衡时刻，无状态日志，故实际接触未知、断点时间留空，原生前向距离也未报告。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点 track_finished 未达到（评分器 replay_success=0，final=0）与帧一致；视频远景中跑者仅占上沿极少像素，接触未知、断点时间留空的处理诚实。现象与根因均已标为推断，我未找到能推翻的证据。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:50:22Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--ifembbr/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--ifembbr/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "5dd964de3a89914b81069ab3d9817f8f67f3038b9924f40508c3759dbe267aae",
+   "video_sha256": "88f832a8e00d06087521aac826a1ec601f5df0a18f3e67c2bd13f270d7ada3a5",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--jvnx5kj.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--jvnx5kj.json
new file mode 100644
index 0000000000000000000000000000000000000000..1cb774b81f3aceb2ce186889b801dd8c2ce7f484
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-privileged--jvnx5kj.json
@@ -0,0 +1,169 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-privileged--jvnx5kj",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 218,
+   "sim_steps": 218,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 360,
+   "agent_resets": 1609,
+   "agent_sim_steps": 68332,
+   "agent_records": 0,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 42.975,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.239,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_memory_gb": 1,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-privileged--jvnx5kj/replay.mp4",
+  "duration_s": 8.76,
+  "fps": 25.0,
+  "frames": 219,
+  "sha256": "eabb870b2c98b9df3b5671a991ee21372e989607a53f3caac34252eef03660fc"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 64,
+      "text_zh": "观察：log step 62 中 agent 的生成脚本保存了218行肌肉控制动作；log step 64 中基准回放工具对该交付文件两次完整回放，均报告 success=false、steps=218、replayed_rows=218，且终态哈希一致、instance_ok=true、deterministic=true。评分器 reward 同样报告 replay_success=0、task_failure=1、sim_steps=218。"
+     },
+     {
+      "t": 8.72,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00–8.72 s 的全程抽帧主要显示固定的橙色跑道；视频 7.76–8.72 s 逐帧检查及画面上沿放大后，仍无法辨认完整跑者、足部支撑或终点到达关系，末段上沿仅露出很小的模型片段。"
+     },
+     {
+      "t": null,
+      "log_step": 64,
+      "text_zh": "推断：该节点对应原生 solved 成功事件，首次满足便结束回合；完整交付回放明确以任务失败结束且从未报告成功，故“跑到终点”未达成，区间为空、终值为假。此判断依据原生回放结果，不将会话中其他控制器试跑当作节点进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00–8.72 s 未清楚呈现跑者足部与跑道的接触；无状态日志接触对，无法判断左脚、右假肢足端实际如何支撑，故不照抄任务DAG的典型接触方案。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 62,
+   "where": "操作中",
+   "evidence_zh": "推断：log step 61–62 显示 agent 的跑步控制器在冻结初态上执行并保存218行动作，随后基准回放工具在 log step 64 确认该文件完整执行后失败；这是执行“跑到终点”时未完成的单一步骤。视频 0.00–8.72 s 未提供可辨认的全身运动，无法定位发生偏差的视频时刻，故 t=null。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "观察：inputs/ 有完整视频与 run.json，但没有状态日志或成功参考视频。视频 0.00–8.72 s 的全程抽帧、上沿裁剪以及视频 7.76–8.72 s 逐帧检查均不能看清跑者全身和足地接触，不能据此区分跌倒、越界等具体物理现象，也无法给出失败的视频时间。log step 62 的生成脚本诊断在仿真 1.00 s 打印 z=0.34、pitch=61.0；log step 68 中 agent 自述左腿单支撑时前倾跌倒，但基准回放工具在 log step 64 仅给出“fallen, off the track or out of time”的合并原因，未报告具体失败分支。会话诊断与 agent 的解释缺乏可见姿态或状态日志的交叉核对，因此现象和根因留空；不把其他试跑的力矩误差、已修复代码错误当成本次物理失败的确定原因。评分器未报告 forward_distance_m；log step 62 的 prog 是 agent 脚本自算值，不作为原生指标。未建立可核对的仿真时钟到视频时钟映射。"
+ },
+ "summary_zh": "观察：交付的218行动作经两次完整确定性回放均失败（log step 62、64），唯一目标“跑到终点”未达成，最终节点进度为0/1。推断：断点位于该行进步骤的操作中；视频主要显示跑道而未清楚呈现跑者姿态或接触，无法核实 agent 在 log step 68 自述的跌倒机制，故接触记未知，现象、根因及断点视频时间留空。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "视频全程只显示橙色跑道，看不到跑者，无法判定节点接触和失败机制；标签把 track_finished 记为未达成（依据 log step 64 的基准回放两次均 failure），断点、未知接触和空的现象/根因层（附 gap_zh）都有依据，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:46:39Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--jvnx5kj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-privileged--jvnx5kj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "fcc4a7bc36c6c265dafe8725f9a3e59db10289fafe8b97d05fdae94aac1c6943",
+   "video_sha256": "eabb870b2c98b9df3b5671a991ee21372e989607a53f3caac34252eef03660fc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--hkjamwl.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--hkjamwl.json
new file mode 100644
index 0000000000000000000000000000000000000000..efe090284fd06babc0bef025de49ffd17346f9ea
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--hkjamwl.json
@@ -0,0 +1,202 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-standard--hkjamwl",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 186,
+   "sim_steps": 186,
+   "physics_steps": 930,
+   "steps": 1568,
+   "calls": 134,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 1568,
+   "elapsed_s": 1564.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.564,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.179,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-standard--hkjamwl/replay.mp4",
+  "duration_s": 7.48,
+  "fps": 25.0,
+  "frames": 187,
+  "sha256": "c8b5eae75737f6a7f05a294f5843aa5cc30e8c86b89ace0692e1a28217b8a6eb"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 77,
+      "text_zh": "观察：log step 76 的脚本重置服务后执行肌肉命令；log step 77 服务返回 episode=11、resets_left=0、control_steps=20、sim_time_s=0.1，之后直到 log step 121 均为同一回合且没有再次重置。任务文本规定仅最后一回合受评。"
+     },
+     {
+      "t": null,
+      "log_step": 120,
+      "text_zh": "观察：log step 120 的第11回合服务结果为 done=True、reason=\"failure: fallen, off the track or out of time\"、control_steps=186、仿真 0.93 s；log step 121 再次查询仍为失败终态。"
+     },
+     {
+      "t": null,
+      "log_step": 120,
+      "text_zh": "推断：原生成功事件一旦触发即结束该回合，而第11回合从开始持续运行至失败终态，故“跑到终点”事件未发生，intervals=[]、final=false；此结论依服务终止结果及事件规则，不仅依据 success=0。"
+     },
+     {
+      "t": 7.44,
+      "log_step": null,
+      "text_zh": "观察：视频 0–7.44 s 的全程粗采样、全部21种解码画面和视频 6.44–7.44 s 的逐帧检查均只显示固定跑道与标志，没有可辨认的跑者，不能从画面核对身体前移、触地或跌倒；本节点据最后一回合服务记录判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 0–7.44 s 未显示跑者及脚部；第11回合日志记录肌肉命令与本体感觉，但无足端接触对或地面反力，无法确认左脚、右侧假肢足端与跑道的实际支撑接触。log step 64 的接触打印来自智能体另建的本地模型，不能用于本评分回合。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 103,
+   "where": "操作中",
+   "evidence_zh": "推断：第11回合正在执行向前行进的肌肉控制，log step 103 服务在仿真 0.65 s 返回 root_z=0.731 m、root_vz=-1.169 m/s；log step 110 在仿真 0.75 s 返回 root_z=0.573 m；log step 120 在仿真 0.93 s 返回 root_z=0.151 m 并判失败，持续下坠未恢复，断在“跑到终点”的操作中。视频不显示身体，无法定位视频时刻，t 留空。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": null,
+    "log_step": 120,
+    "evidence_zh": "推断：第11回合服务本体感觉显示骨盆从仿真 0.65 s 的0.731 m（log step 103），降至仿真 0.85 s 的0.362 m（log step 114）、仿真 0.90 s 的0.227 m（log step 116），最终仿真 0.93 s 仅0.151 m且竖直速度为-2.487 m/s，随后服务判失败（log step 120）。连续高度坍落结合失败终态支持失衡跌倒；不是仅凭智能体最终自述“跌倒”。视频中身体不可见，无法确认具体倒向或着地部位。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 106,
+    "t": null,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 103、106、109 的脚本在同一评分回合依次改变双侧髋屈肌、膝肌与侧向稳定肌命令，尝试把腿收回并恢复支承；服务返回骨盆高度在仿真 0.65、0.70、0.75 s 依次为0.731、0.661、0.573 m。log step 115 又提高左膝伸肌命令至0.9，但 log step 116、120 的骨盆继续下降至0.227、0.151 m并终止。肌肉协调与恢复控制未能抑制下坠，支持笼统的控制归因；无足端受力或可验证的稳定步态，不能进一步认定某个增益、肌肉方向或接触设计是根因。"
+   }
+  ],
+  "gap_zh": "视频存在，但视频 0–7.44 s 全程只见固定跑道，没有跑者；最后一秒已逐帧检查。无状态日志、无成功参考视频，第11回合服务日志也未给出接触对或独立的跌倒分支字段，平衡现象由骨盆连续坍落与失败终态推断。服务记录的仿真终点为0.93 s，视频长度为7.48 s，无法用可见事件核对二者映射，故不把仿真时间写作视频时间，断点与现象的 t 均留空。实际接触未知，控制根因不细分；评分器未报告 forward_distance_m，原生指标值留空。"
+ },
+ "summary_zh": "最后受评的是第11回合，智能体已执行行进肌肉控制，但“跑到终点”未达成，最终为0/1。服务记录中骨盆在仿真 0.65–0.93 s 持续下降并以失败终止，支持行进中失衡跌倒及控制未能恢复的归因。视频全程仅见固定跑道，实际接触和失败的视频时刻无法核实，均保留未知。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频全程只见固定跑道、无跑者，label 据最后一回合服务记录判未达终点、断点留空 t、平衡+控制归因，均有服务读数支持；时间未转换，未发现需改内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:37Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--hkjamwl/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--hkjamwl/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "212a2a99645dd3a41a47df2bed6f88ad618932ce8f06d44e7421497a1e5858e9",
+   "video_sha256": "c8b5eae75737f6a7f05a294f5843aa5cc30e8c86b89ace0692e1a28217b8a6eb",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--kayadpw.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--kayadpw.json
new file mode 100644
index 0000000000000000000000000000000000000000..ff672811f6243ccb18815814eaf5ffce33d763ad
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--kayadpw.json
@@ -0,0 +1,180 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-standard--kayadpw",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 291,
+   "sim_steps": 291,
+   "physics_steps": 1455,
+   "steps": 623,
+   "calls": 41,
+   "episodes": 4,
+   "resets": 3,
+   "reset_budget": 10,
+   "sim_steps_total": 623,
+   "elapsed_s": 3084.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 4.057,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.584,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-standard--kayadpw/replay.mp4",
+  "duration_s": 11.68,
+  "fps": 25.0,
+  "frames": 292,
+  "sha256": "3be97562b8e3a3a01ce37d4bad8a5a99c5d5725f74b97c3f087f9821712eb439"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 389,
+      "text_zh": "观察：log step 388 启动交付脚本；log step 389 的服务状态显示 episode=4、control_steps=0 后开始提交动作，随后多批动作返回 done=False。脚本打印的 dist 与 root 坐标来自本地并行仿真（log step 328 的代码），不是服务端实际位姿，故不据此标记距离进展。"
+     },
+     {
+      "t": 11.64,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0–11.5 s 按 0.5 s 间隔检查，视频 10.64–11.64 s 逐帧检查，画面均只见橙色跑道及黑色背景，未见跑者、足部接触或到达终点的动作；该视角无法直接判断行进状态。"
+     },
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "推断：inputs/run.json 的 reward 报告 graded_by_live=1、episodes=4、n_actions=291、finalized=1、live_success=0、task_failure=1，且 infra_error=sim_error=grader_error=0。DAG 的唯一目标按原生 solved 首次触发即成功结束；因此本次最后在线回合未触发跑到终点事件，final=false、intervals=[]。这不是仅由零奖励推断未尝试。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：log step 389 表明脚本已为最后评分回合提交行进动作，但视频 0.0–11.64 s 的已检查画面不显示跑者，且无状态日志，无法核实左脚或右侧假肢足端与地面的实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 389,
+   "where": "操作中",
+   "evidence_zh": "推断：log step 388–389 显示 agent 的交付脚本重置到 episode=4 后实际提交行进控制，服务返回该回合尚未结束；inputs/run.json 的最终评分记录该回合执行291个动作后 task_failure=1、live_success=0。因此断点为唯一行进步骤“跑到终点”的操作中。视频 0.0–11.64 s 的已检查画面均不显示跑者，无法定位具体出错时刻。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "观察：inputs/ 有完整视频和运行日志，但无状态日志及成功参考视频。视频 0.0–11.5 s 的每0.5 s抽帧和视频 10.64–11.64 s 的逐帧画面仅显示跑道，跑者及足部接触不可见；最后评分回合的日志反馈止于 log step 389 的运行中状态，没有其终止时的具体失败分支。log step 384 的失败状态属于 episode=3，不能代替最后的 episode=4；log step 394 的1.57米是 agent 自述，未由评分器距离字段或可见动作核实。推断：现有证据能确定未完成行进及已提交动作，但不能区分跌倒、出界或其他物理失败，也不能确定控制失误的具体原因，故现象与根因留空、实际接触标未知、原生前进距离留null。服务200 Hz记录（log step 7）与视频时钟没有可见事件可供映射核对，故不从控制步推定视频失败时刻。"
+ },
+ "summary_zh": "最后评分回合已提交291个动作，但原生检查未报告到达终点，唯一目标节点最终完成0/1，断点为“跑到终点”的操作中。视频全程的已检查画面仅显示跑道，看不到跑者；缺少最后回合具体失败分支及状态日志，无法确认实际接触、失败时刻、现象和根因，评分器也未报告前进距离。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "标签保守但与证据一致：视频只见跑道、无法见跑者；最后回合 episode=4、n_actions=291、task_failure=1，断点在唯一步骤“跑到终点”的操作中；现象与根因因证据无法区分跌倒/出界而留空并有 gap_zh，合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:48:59Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--kayadpw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--kayadpw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "14e622b494493f0ee83161036a3f2e271df79147218124de5f3f0b4a68c5b51e",
+   "video_sha256": "3be97562b8e3a3a01ce37d4bad8a5a99c5d5725f74b97c3f087f9821712eb439",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--mzbpnvx.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--mzbpnvx.json
new file mode 100644
index 0000000000000000000000000000000000000000..a96e574a7348c289977c44d740f29e3913025c7b
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--mzbpnvx.json
@@ -0,0 +1,195 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-standard--mzbpnvx",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 1,
+   "sim_steps": 1,
+   "physics_steps": 5,
+   "steps": 1,
+   "calls": 4,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 1,
+   "elapsed_s": 3731.9,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.833,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.384,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-standard--mzbpnvx/replay.mp4",
+  "duration_s": 0.08,
+  "fps": 25.0,
+  "frames": 2,
+  "sha256": "fca1929abb219b6c51500c33b802b6f934b3f2f7e655312bf7b054d720d8ebd4"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": 43,
+      "text_zh": "观察：log step 43 中，agent 的脚本仅向正式机器人服务提交一行全零 actuators 动作，dt=0.005；服务返回 control_steps=1、done=false、reason=null。该调用明确用于核对单步仿真机制，并非执行跑向终点的控制序列。"
+     },
+     {
+      "t": 0.04,
+      "log_step": 82,
+      "text_zh": "观察：视频 0.00 s、0.04 s 两帧均只有跑道俯视画面，无法辨认跑者或终点到达动作；完整日志后续至 log step 82 均未再提交正式动作，末条 say 为空。评分器记录 n_actions=1、sim_steps=1、episodes=1、resets=0、live_success=0、replay_success=0、timeout=1，未报告前向距离。"
+     },
+     {
+      "t": 0.04,
+      "log_step": 43,
+      "text_zh": "推断：结合正式服务仅执行一个诊断控制步且尚未结束、之后没有继续推进该回合，以及最终原生成功结果为零，跑到29米终点的事件从未达成；此判断不采用本地测试的位移或终止读数，也不把 agent 工作时间映射为视频时间。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 0.00 s、0.04 s 无法辨认跑者和足端；log step 43 的正式单步诊断没有足地接触读数，无法判断左脚或右侧假肢足端实际如何接触跑道。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": 0.0,
+   "log_step": 43,
+   "where": "首步前",
+   "evidence_zh": "推断：log step 43 仅执行用于辨识服务单步机制的全零诊断，服务报告 control_steps=1、done=false；log step 49–82 的步态开发与测试没有后续正式 submit，评分器最终仍记 n_actions=1、sim_steps=1。有效跑向终点的步态在受评回合未启动，因此断在“跑到终点”首步前；视频 0.00 s 仅作导航锚点，画面不能定位物理失败时刻。"
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 43,
+    "evidence_zh": "推断：log step 43 的脚本只提交单行全零诊断，返回 control_steps=1、done=false；其后直到 log step 82 只在本地仿真中开发、测试和读取源码，评分器的正式动作总数仍为1。跑29米的控制序列从未在受评回合执行；视频 0.00 s、0.04 s 均无可辨认的跑步动作，不能将会话测试中的失稳记为本回合跌倒。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 68,
+    "evidence_zh": "推断：agent 的本地控制器未形成可持续行进：log step 53 在本地 gait tracking 测试计数200处输出骨盆高度0.20 m、滚转约−72.5°且 done=1；log step 60 的另一状态机在本地测试计数137处输出高度0.28 m、俯仰约−63.6°且 done=1；log step 68 的步态加平衡反馈在本地测试计数209处输出高度0.26 m、滚转约−76.8°且 done=1。代码均推进本地 gym 环境，并未向正式服务提交这些序列；结合正式回合仅保留 log step 43 的诊断，支持控制实现未能产出可用跑步序列这一根因，尚不足以细分具体控制缺陷。上述本地测试计数均不是视频时间。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "已检查 inputs/replay.mp4 的全部两帧（视频 0.00 s、0.04 s）；只有跑道俯视画面，跑者、足地接触及实际位移不可辨认。输入没有状态日志、成功参考视频，也没有评分器前向距离值，因此接触记未知，原生距离指标留空；正式单步诊断与原生成功记录足以判断终点未达。根因仅由本地多种控制器反复失稳推断，日志没有最终说明为何未继续提交，也无法确定具体是接触设计、轨迹生成还是反馈控制问题；本地测试不用于受评回合的物理现象或节点进度。",
+  "unmet": {
+   "track_finished": false
+  }
+ },
+ "summary_zh": "正式回合仅执行一个用于核对仿真机制的全零诊断控制步，未启动跑向29米终点的有效步态，节点最终完成0/1，断点为“跑到终点”的首步前。会话中多种本地步态控制器反复失稳，支持控制能力不足的推断；这些测试未进入受评轨迹。视频仅有两帧跑道俯视图，无法辨认跑者和实际足地接触，且没有状态日志或前向距离报告。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与证据一致：正式服务只提交了 log step 43 的一步全零诊断（control_steps=1、n_actions=1），跑向29米终点的步态未在受评回合启动，断点 track_finished / 首步前 / 没尝试，视频仅两帧跑道俯视图，接触记未知合理。超时未被当作原因，根因控制源自本地多控制器反复失稳（会话证据，参照 FAIL-22），已标明推断，且 gap_zh 说明不足。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:54:36Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--mzbpnvx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--mzbpnvx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "06f468dad9d68e39a5290eac5d124baaaa0dff3f862ad70a58060581c7e0e793",
+   "video_sha256": "fca1929abb219b6c51500c33b802b6f934b3f2f7e655312bf7b054d720d8ebd4",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--qmh6snf.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--qmh6snf.json
new file mode 100644
index 0000000000000000000000000000000000000000..6009bb6c1d8ff1da3eb5d150fc284940b6073c5d
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--qmh6snf.json
@@ -0,0 +1,184 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-standard--qmh6snf",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 449,
+   "sim_steps": 449,
+   "physics_steps": 2245,
+   "steps": 1009,
+   "calls": 1096,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 2697,
+   "elapsed_s": 2937.9,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-standard--qmh6snf/replay.mp4",
+  "duration_s": 18.0,
+  "fps": 25.0,
+  "frames": 450,
+  "sha256": "2dcd876b27211675efe6c4ca2cc101cf50a3c15a60101247cbc959ab2e60cd25"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 106,
+      "text_zh": "观察：log step 106 的脚本执行最后一次 reset，回放早期回合的开场肌肉命令后持续在 episode 11 执行跑步控制；log step 130 的服务状态确认 episode=11、resets_left=0、control_steps=449。这是评分采用的最后一回合，先前被重置的回合不计进度。"
+     },
+     {
+      "t": null,
+      "log_step": 130,
+      "text_zh": "推断：log step 130 的服务终态为 done=true、reason=\"failure: fallen, off the track or out of time\"，评分器 live_success=0、replay_success=0、live_matches_replay=1。按本任务首次 solved 即结束回合的规则，最后一回合没有触发跑到终点事件，故本节点全程未达成；此判断依据服务终态与验收规则，不由零分单独推出。"
+     },
+     {
+      "t": 17.96,
+      "log_step": null,
+      "text_zh": "观察：视频 0–17 s 每秒抽查及视频 17.00–17.96 s 逐帧检查均仅见跑道，未见可辨认的身体、脚或距离标尺；画面不能核对身体移动、足地接触或到达终点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：最后一回合的跑步命令见 log step 106、109–127，但所检视频帧只有跑道，未显示脚、假肢或身体与地面的接触；没有状态接触日志，无法确认实际作用端及支撑接触，不复制任务 DAG 的典型接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 123,
+   "where": "操作中",
+   "evidence_zh": "推断：最后一回合一直执行向终点行进及恢复支撑的命令（log step 106、116、121、125、127），未完成跑到终点。log step 123 的服务本体读数显示骨盆高度在仿真 1.9 s、2.0 s、2.1 s 依次为 0.672、0.423、0.191 m，仿真 2.0 s 竖直速度为 −3.000 m/s；随后恢复命令未使其重新站起，log step 129 在仿真 2.245 s 返回 done=true，log step 130 确认失败。故断点为行进操作中的支撑失稳；视频未显示身体，不能给出经过可见事件核对的视频时间。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": null,
+    "log_step": 123,
+    "evidence_zh": "推断：episode 11 的服务本体读数在 log step 119 尚有仿真 1.7 s 骨盆高度 0.809 m；log step 123 显示仿真 2.1 s 高度降到 0.191 m，log step 126、129 的后续读数持续约 0.21 m，末次骨盆俯仰约 −0.898 rad 且服务随后判负。身体迅速下落并未恢复站立支持平衡失败，不能把服务的合并失败文案当作已明确报告 fallen 分支；该现象来自服务读数的推断，视频没有显示倒地。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": null,
+    "log_step": 121,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 116 的脚本在 episode 11 改变支撑相位以恢复身体，log step 121 又针对 pitch<−0.15 增加髋、膝关节力矩修正，log step 122 展示其用本体姿态和速度计算目标力矩并招募肌肉的控制代码；但 log step 123、126、129 的服务读数显示骨盆仍从仿真 1.7 s 的 0.809 m 降至约 0.21 m，恢复肌肉命令执行后也未重建站立。因此证据支持该步的肌肉与支撑控制未能稳定身体；未证明某个特定增益、肌肉映射或接触方案是唯一原因，停留在控制大类。"
+   }
+  ],
+  "gap_zh": "视频存在，但视频 0–17 s 每秒抽查和视频 17.00–17.96 s 逐帧检查只见跑道，没有可辨认的身体或足地接触；无 states.jsonl、无其他可用视角及成功参考视频。log step 106 的最后一次 reset 与 log step 130 的 episode=11 可界定最终评分回合，服务本体读数支持支撑失稳的推断，但无法在视频核对任何身体事件，因此不把仿真秒、agent 工作时间或 control_steps 换算成视频时间，breakpoint.t 和现象 t 留空。原生失败原因合并多个分支，不能核实具体判负分支；接触标未知，控制子类和具体参数原因不作推定。评分器没有报告 forward_distance_m，指标值留空。",
+  "unmet": {
+   "track_finished": false
+  }
+ },
+ "summary_zh": "最后评分回合为 episode 11，跑到终点节点未达成，最终进度为 0/1；此前 10 次重置的回合不计入。最后一回合的服务本体读数显示骨盆快速下降至约 0.2 m，恢复命令后仍未站起并最终判负，支持平衡失败及控制不足的推断。视频只有跑道，无法核对身体、实际接触和失败的视频时刻，原生前向距离也未报告。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "视频全程只有跑道、不见身体，标签未强行换算视频时间，节点未达成与最终0/1有服务终态（episode 11, done=true failure）支持；骨盆高度下降的服务读数支持平衡失败的推断，控制根因停在大类，标注诚实。未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:54:54Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--qmh6snf/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--qmh6snf/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a6bd59ff7b8c2edd49f8df4e2ba344098f7b4eed9fd04f85af401f99b9d9f748",
+   "video_sha256": "2dcd876b27211675efe6c4ca2cc101cf50a3c15a60101247cbc959ab2e60cd25",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--sbzftfh.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--sbzftfh.json
new file mode 100644
index 0000000000000000000000000000000000000000..58047716e8c2d7635f1b8edd0fcf8aad8ad4b10d
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--sbzftfh.json
@@ -0,0 +1,196 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-standard--sbzftfh",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 382,
+   "sim_steps": 382,
+   "physics_steps": 1910,
+   "steps": 1634,
+   "calls": 51,
+   "episodes": 5,
+   "resets": 4,
+   "reset_budget": 10,
+   "sim_steps_total": 1634,
+   "elapsed_s": 2871.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.65,
+   "peak_processes": 16,
+   "peak_disk_gb": 0.456,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-standard--sbzftfh/replay.mp4",
+  "duration_s": 15.32,
+  "fps": 25.0,
+  "frames": 383,
+  "sha256": "d7e69dc05b4803866936d46d5c14e7068a6a80528f6f2c5ea370777088415567"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：已按视频 0–15.28 s 的半秒帧及视频 12–15.28 s 的密集帧检查全程。固定远景中肌骨模型从画面上缘进入，仅在跑道起端附近移动，始终没有穿过整条跑道到达另一端；视频 15.28 s 仍在画面上缘附近。"
+     },
+     {
+      "t": 15.28,
+      "log_step": 75,
+      "text_zh": "观察：log step 75 的服务状态明确记录最后回合 episode=5、control_steps=382、done=True，终止原因为合并提示“failure: fallen, off the track or out of time”，并打印服务本体感知的 root_y=12.901262；评分器字段 live_success=0、task_failure=1。该回合控制步数与评分器 n_actions=382 一致，前四回合不计入本次节点进度。"
+     },
+     {
+      "t": 15.28,
+      "log_step": 75,
+      "text_zh": "推断：log step 12 记录冻结起点 root_y=15，log step 75 重置后报告 start diff=0.0，因此最后回合前移约 15−12.901262=2.10 米，远未达到29米；结合视频中模型仅在起端附近移动，判定节点从未达成。此距离是从服务本体感知推算，并非评分器直接报告的原生距离指标。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 3–15.28 s 可见模型在跑道上移动及轮廓变化，但模型只有数个像素，起始部分被画面上缘裁切；放大后仍无法分辨左脚、右侧假肢足端或躯干与地面的具体接触，不能将DAG的典型交替足端支撑直接当作已观察接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": 14.0,
+   "log_step": 75,
+   "where": "操作中",
+   "evidence_zh": "推断：机器人正在执行跑向终点的行进动作，视频 14.0–15.28 s 的小模型轮廓展开并改变朝向，随后视频结束；log step 75 证实第5回合因原生失败条件终止，前移仅约2.10米。结合 agent 在 log step 76 自述最后回合跌倒，断点归于“跑到终点”操作中失衡，未在其后修复；视频 14.0 s 仅为末段失衡的粗略导航锚点，无法精确定位首次失去支撑的帧。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 14.0,
+    "log_step": 75,
+    "evidence_zh": "推断：视频 14.0–15.28 s 放大帧中的模型轮廓由较紧凑变为展开、转向，位置仍在跑道起端附近；log step 75 服务返回原生失败的合并提示，agent 在 log step 76 自述第5回合前移约2.10米后跌倒。合并这些证据判断为行进中的失衡跌倒；远景不足以独立辨认身体落地部位，合并提示本身也不单独证明跌倒。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 76,
+    "source_zh": "日志",
+    "evidence_zh": "推断：agent 的脚本采用参考步态关节跟踪并将目标力矩转换为肌肉命令，另加骨盆姿态与支撑侧修正（log step 21、31、51）；最终用该控制器及调参结果执行第5回合仍在短距离后终止（log step 75）。会话中的独立站立测试在提高关节刚度后仍出现整体倾斜和骨盆高度下降（log step 48–49；仅作根因辅助证据），agent 在 log step 76 也自述未找到稳定步行控制。由此归为控制能力不足；没有足够证据锁定某个增益、接触布局或反馈环节，不细分子类。"
+   }
+  ],
+  "gap_zh": "现有视频为320×240固定远景，模型在上缘仅占数个像素，部分起始动作出画，无法独立核实足端接触、首次失衡时刻及最终着地部位；没有状态日志或成功参考视频。最后回合服务仅给跌倒/出界/时间的合并失败提示，因此跌倒现象依赖末段轮廓变化与 agent 自述的联合推断。382个控制步按任务说明对应仿真1.91 s，而视频长15.32 s；未建立逐步时钟映射，视频锚点均直接取所检帧。评分器未报告 forward_distance_m，原生指标保留null；具体控制失稳机制也未得到验证。",
+  "unmet": {
+   "track_finished": false
+  }
+ },
+ "summary_zh": "评分对象为最后的第5回合，“跑到终点”从未达成，最终0/1；服务本体感知结合冻结起点推算前移约2.10米，远未达到29米。末段画面与 agent 自述联合支持行进中失衡跌倒，控制器及会话测试支持宽泛的控制根因。视频中模型过小且无状态日志，实际接触标为未知，跌倒起始时刻和具体控制机制仍有证据缺口。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成（0/1）、断点 track_finished、平衡现象与控制根因均与服务读数（log step 75：episode 5、control_steps=382、done、root_y=12.90）及视频一致；视频中模型仅数像素，接触标未知合理，推断均已标注。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:47:32Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--sbzftfh/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--sbzftfh/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3c5443c78efc39b4a25f15372c6c7510c129e4ef887746c5cd975c158906dad9",
+   "video_sha256": "d7e69dc05b4803866936d46d5c14e7068a6a80528f6f2c5ea370777088415567",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--ukth7qv.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--ukth7qv.json
new file mode 100644
index 0000000000000000000000000000000000000000..3aea2bc8e33313881f3dc3c210df891d281718f7
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--ukth7qv.json
@@ -0,0 +1,188 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-standard--ukth7qv",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 215,
+   "sim_steps": 215,
+   "physics_steps": 1075,
+   "steps": 1534,
+   "calls": 75,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 1534,
+   "elapsed_s": 1869.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.239,
+   "peak_processes": 23,
+   "peak_disk_gb": 1.21,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-standard--ukth7qv/replay.mp4",
+  "duration_s": 8.64,
+  "fps": 25.0,
+  "frames": 216,
+  "sha256": "986f76c78ef82b7a57cd9888c8822233cb1ddefae1c13b472c1eca7e9263b8db"
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "log",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 153,
+      "text_zh": "观察：最后一次服务执行返回 episode=11、control_steps=215、done=True，原因为“failure: fallen, off the track or out of time”；该回合此后未被重置。评分器 reward 的 episodes=11、n_actions=215、success=0、live_success=0 与此吻合。"
+     },
+     {
+      "t": null,
+      "log_step": 153,
+      "text_zh": "推断：任务以首次原生成功事件立即结束回合，最后回合却以失败结束，因此“跑到终点”从未成立；并非仅凭零分推断未尝试。agent 脚本打印的服务本体感知终值为 root_y=13.93 m、root_z=0.10 m，并按其起始 root_y=15 m 计算前移约1.07 m，远小于29 m。"
+     },
+     {
+      "t": 8.6,
+      "log_step": null,
+      "text_zh": "观察：已按视频0.0–8.5 s每0.5 s查看全程，并逐帧检查视频7.60–8.60 s；画面只有橙色跑道、灰色边缘和标志，未见跑者，无法从视频观察奔跑、足地接触或到达终点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：log step 153记录agent脚本对最后回合提交肌肉控制命令，故该步骤已尝试；所查视频画面不含跑者，且无状态日志，不能确认左脚、右假肢足端或躯干实际如何接触跑道。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": null,
+   "log_step": 153,
+   "where": "操作中",
+   "evidence_zh": "观察：log step 153中，agent脚本在最后第11回合执行步态控制，服务本体感知经脚本打印为骨盆高度依次约0.39、0.25、0.11、0.10 m，随后服务返回done=True和合并失败原因；该回合215个控制步与评分器记录一致。“跑到终点”未完成。视频中不见身体，不能定位失稳的视频时刻。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": null,
+    "log_step": 153,
+    "evidence_zh": "推断：log step 153明确来自被评分的第11回合；agent脚本读取服务返回的本体感知，打印骨盆root_z由0.39 m连续降到0.25、0.11、0.10 m，随后服务判负，支持身体塌落、失去站立平衡的判断。此证据是该回合的服务读数，不是log step 154中agent关于假肢膝失稳的自述；视频不含跑者，不能直接目视确认跌倒姿态。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "观察：inputs/有视频和完整会话日志，但没有状态日志或成功参考视频；所查全程及末秒逐帧画面均未见跑者，实际接触记未知。服务仅给出合并失败原因，平衡现象依据最后回合的骨盆高度下降推断，未获得原生跌倒分支的独立确认。log step 154中agent自述假肢膝在落地负荷下屈曲，但未提供最后回合的膝角、负荷、目标与跟踪误差，不能确认具体控制根因，故根因留空。215个控制步按任务200 Hz为仿真1.075 s，视频却长8.64 s且无可核对的动作事件，不建立两钟映射，断点视频时刻留空；脚本打印的t为执行墙钟耗时，不用作视频时间。评分器reward未报告前向距离，native_metrics的value保留null，日志脚本计算的约1.07 m仅作补充证据。"
+ },
+ "summary_zh": "最后第11回合已执行奔跑控制，但未到达29米终点，节点完成0/1；log step 153中agent脚本由服务本体感知计算前移约1.07 m后服务判负。该回合骨盆高度降至约0.10 m，支持在“跑到终点”的操作中失衡的推断；视频全程所查画面只见跑道而不见身体，无法确认实际接触、跌倒的视频时刻或具体控制根因。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标注与证据一致：track_finished 未达成（服务返回 done=True、control_steps=215、评分器 success=0），视频只有跑道不见跑者，故接触未知、时间留空合理；平衡现象基于最后回合骨盆高度降至0.10 m 的服务读数，标为推断且说明局限；根因留空并写明 gap_zh，没有未经支持的钟换算。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T03:53:21Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--ukth7qv/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--ukth7qv/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "dcc8d0c6fcdd2c7066330de37553288f311a8b80cc560721ac90b4e435bf0265",
+   "video_sha256": "986f76c78ef82b7a57cd9888c8822233cb1ddefae1c13b472c1eca7e9263b8db",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--z9aamsa.json b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--z9aamsa.json
new file mode 100644
index 0000000000000000000000000000000000000000..5848ebf9cf90d1e4e613619fc1e76db7c8f0d2f2
--- /dev/null
+++ b/phase2/myosuite/osl_run/trajectories/myosuite-osl-run-i00-standard--z9aamsa.json
@@ -0,0 +1,218 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-osl-run-i00-standard--z9aamsa",
+ "taskKey": "myosuite/osl_run",
+ "dag_sha256": "e5e4a4737e99dbffcc27d3a2017c662faacbdb962fff22a2d9a6f2165acb94fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 29510,
+   "sim_steps": 29510,
+   "physics_steps": 147550,
+   "steps": 71748,
+   "calls": 369,
+   "episodes": 6,
+   "resets": 5,
+   "reset_budget": 10,
+   "sim_steps_total": 71748,
+   "elapsed_s": 3008.3,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.81,
+   "peak_processes": 16,
+   "peak_disk_gb": 1.196,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-osl-run-i00-standard--z9aamsa/replay.mp4",
+  "duration_s": 59.12,
+  "fps": 25.0,
+  "frames": 1478,
+  "sha256": "828fe600be05d1df66f0fb086b2dbd48d9b1ede9e9cfea40a29960db3bba1bff"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "track_finished": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 0–3 s 的密集抽帧及全片每约 4 s 的抽帧中，模型只在画面上缘的跑道远端占据少数像素，始终停留在这一小片区域，没有沿整条可见跑道通过；这些帧不足以辨认具体肢段动作。"
+     },
+     {
+      "t": 59.08,
+      "log_step": 188,
+      "text_zh": "观察：视频 58.0–59.08 s 的密集帧中，模型仍在画面上缘远端，未通过跑道；log step 188 的服务最终状态为 episode=6、control_steps=29510、done=true、reason=\"failure: fallen, off the track or out of time\"，与评分器 live_success=0 一致。"
+     },
+     {
+      "t": null,
+      "log_step": 174,
+      "text_zh": "观察：log step 174 启动 best_p2 控制器并重置到 episode=6；该回合 c22–c24 的服务本体读数为骨盆 y=14.12–14.13 m、z=0.32–0.33 m，log step 188 的 c147 读数为 y=13.50 m、z=0.46 m。推断：结合视频中模型始终停在跑道远端，以及 log step 188 服务 done=true 和评分器 live_success=0，最后回合未跑到终点；此前已重置的回合和本地孪生测试不计入节点进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "推断：视频约 1–57 s 模型一直停留在跑道远端，结合 log step 174、178、182 的低骨盆高度读数，实际承重交互为下肢/本体与跑道地面的支撑接触；远景和画面上缘裁切不足以分清左足、假肢足端或其他腿段分别何时着地，不能确认正常交替跑步。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "forward_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 29,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "track_finished",
+   "t": 1.0,
+   "log_step": 174,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 1 s 及后续全片抽帧中，模型始终位于画面上缘的跑道远端，至视频 59.08 s 仍未通过跑道，帧不足以分辨肢段动作；log step 174、178、182 的 episode=6 服务本体读数中，骨盆 z 多次为 0.32–0.33 m，log step 188 的 c145 为 0.36 m。推断：断点是正在执行的“跑到终点”，本体在回合早期失去正常站立支撑后未恢复有效行进；视频约 1 s 保留为早期停留的粗略导航锚点，不是由日志换算或精确确认的跌倒时刻。"
+  },
+  "unmet": {
+   "track_finished": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 1.0,
+    "log_step": 174,
+    "evidence_zh": "观察：log step 111 部署脚本在每次服务提交后读取 res.meta.proprio.now，并将其中 [177:180] 打印为骨盆 root=(x,y,z)，不是本地预测值。最后评分回合 episode=6 在 log step 174 的 c22–c24、log step 176 的 c34、log step 178 的 c109–c110、log step 182 的 c115–c116 读数中 z=0.32–0.33 m，log step 188 的 c145 为 0.36 m；此前回合的起始高度参照为 log step 120 reset z=0.864 m、ep3 c0 z=0.79 m，后者仅作站立高度参照。视频 0.48、5、15、30、45、58.48、59.08 s 的远景中，模型始终在画面上缘，无法直接确认姿态。推断：最后回合多次相隔很远的低高度读数支持本体早期塌陷、失去正常站立支撑后未恢复，应标平衡；这些离散读数不证明中间每一时刻高度恒定。log step 188 的结束原因是合并字符串“failure: fallen, off the track or out of time”，不据此单独认定原生跌倒分支。视频约 1 s 只是粗略导航锚点，未将脚本块号或仿真时间换算为视频时间。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 111,
+    "source_zh": "日志",
+    "evidence_zh": "推断：肌肉控制器未能维持站立支撑并形成可持续推进的行进控制。log step 111 的部署脚本使用本地模型预测每个动作块，再向服务提交；log step 174 确认最后回合实际使用 best_p2、1 s 动作块。相同参数在 log step 156 的本地测试延长至仿真 20 s 时也仅前进约 1.35 m，log step 168 的长测试在仿真约 45.3 s 报 fall；这些是会话测试证据。最后回合 log step 174、178、182 的低骨盆高度读数、log step 188 的小幅前向位置变化及视频约 1–57 s 的停留支持控制能力不足，尚不足以锁定某个增益、接触设计或反馈子类。"
+   }
+  ],
+  "gap_zh": "视频只有 320×240 的跑道远景，模型在画面上缘且部分被裁切；已检查全片粗抽帧、开头及视频约 57–59.08 s 的密集帧，并放大视频 0.48、5、15、30、45、58.48、59.08 s 的上缘区域，仍不能可靠辨认逐足接触、塌陷姿态和支撑丢失的精确视频时刻。平衡现象据最后第 6 回合服务本体读数推断，而非仅凭帧或 agent 自述；log step 120 的前一回合仅提供站立高度参照，不计入评分回合进度。inputs/ 无状态日志和成功参考视频；服务只给合并失败原因，未给原生失败分支，因此不确认最终到底触发跌倒、出界还是其他原生结束条件。评分器未报告 forward_distance_m，原生指标保留 null；日志中的本体坐标不冒充评分器指标，亦未把 agent 时间或脚本块计时换算为视频时间。"
+ },
+ "summary_zh": "评分的最后第 6 回合尝试了下肢接地行进，但始终停在跑道起点附近，“跑到终点”未达到，节点终态为 0/1。最后回合的服务骨盆高度多次约为 0.32–0.36 m，支持早期失去正常站立支撑后未恢复的平衡现象，根因仍为控制能力不足，具体控制子类未定。置信度维持 6：视频远景和裁切无法直接确认塌陷姿态及精确发生时刻，服务结束原因未区分原生失败分支，原生距离指标未报告。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点 track_finished=false、断点 track_finished/操作中、控制根因基本可接受；但现象标成“步态（无跌倒的行进太慢）”与服务读数矛盾：最后回合骨盆高度从约 0.79 m 降到约 0.33 m 并长期不变，是失去支撑/倒地而非站立行走缓慢，应记平衡。视频只有远景，帧不能反驳也不能单独证明。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点 track_finished=false、断点（track_finished/操作中）、现象 全身·平衡（据最后回合服务读数 z≈0.32–0.36 m，标为推断）及控制根因均有证据支持，上一轮问题已修复，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:12:30Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--z9aamsa/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--z9aamsa/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--z9aamsa/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-osl-run-i00-standard--z9aamsa/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f056913b6f0a8f0bb8cdc9fb982ad6c2a30bb67e5a591744f3002814c18c4633",
+   "video_sha256": "828fe600be05d1df66f0fb086b2dbd48d9b1ede9e9cfea40a29960db3bba1bff",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--22mvhvw.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--22mvhvw.json
new file mode 100644
index 0000000000000000000000000000000000000000..78471691ec1cb4d92a4987f0a1e9580c21e9b1f4
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--22mvhvw.json
@@ -0,0 +1,190 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-privileged--22mvhvw",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 350,
+   "sim_steps": 350,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 49,
+   "agent_resets": 6271,
+   "agent_sim_steps": 1137,
+   "agent_records": 0,
+   "agent_replays": 1,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.855,
+   "peak_processes": 17,
+   "peak_disk_gb": 0.243,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-privileged--22mvhvw/replay.mp4",
+  "duration_s": 14.04,
+  "fps": 25.0,
+  "frames": 351,
+  "sha256": "0e4edd08cffc8cf9a1c13944a1c027cddb564056f48417a6b4ab24063be8456d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s，白色足球位于画面右下方起球处，远离左上方的球门，初始未进球。"
+     },
+     {
+      "t": 5.5,
+      "log_step": null,
+      "text_zh": "观察：视频 4.4–5.9 s，MyoLegs 模型在白球旁降至低位并伸腿，白球随后脱离模型，向画面左上方球门方向运动；视频 6.0–12.8 s，球持续经过门前草地，未越过门线。"
+     },
+     {
+      "t": 14.0,
+      "log_step": 45,
+      "text_zh": "观察：视频 13.0、13.48、14.0 s，足球均清楚位于球门线前的草地上，与门线仍有间隔；成功参考视频 8.76 s 的白球则到达两门柱之间的门线处。log step 45 保存并回放最终 350 步轨迹，服务返回 success=False、reason=None；评分器字段 replay_success=0、deterministic=1，与本次视频未进球一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "候选接触名：击打。视频 4.6–5.6 s，模型在低位伸出脚/腿触球，白球随即向球门方向离开；词表没有踢击项，按 S16 使用其他。低分辨率下不能可靠区分左右脚及最初接触的具体身体部位。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "视频 4.4–5.9 s，模型伸腿触球期间，另一侧伸向画面右下方的腿脚接触草地，支撑处于低位的本体；其后本体保持低伏、双腿张开的姿态。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 5.5,
+   "log_step": 45,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 5.5 s，MyoLegs 模型伸腿送出白球，此后球向球门运动；视频 13.0–14.0 s，球仍留在门线前，模型也没有再次追球补踢。log step 45 对保存的 350 步轨迹回放得到 success=False；断点为这次未完成的踢球入门步骤，时间以击球处作粗略导航锚点。"
+  },
+  "unmet": {
+   "ball_in_goal": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 5.5,
+    "log_step": null,
+    "evidence_zh": "观察：视频约 5.5 s，模型将白球送向球门；视频 6.0–14.0 s，球在门前向前移动，但最终仍在门线前，没有进入两门柱之间的球门区域。本次射门的球位置未达到目标；未见守门员或门柱碰撞把球挡回。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 5.5,
+    "log_step": 41,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 41 的最终生成器以肌肉组常量叠加定时脉冲，生成固定的 350 步开环肌肉命令；log step 45 修复索引后保存并实际回放该轨迹，返回 success=False。视频 4.4–5.9 s 显示模型在低位伸腿触球，视频 13.0–14.0 s 的球仍在门线前。目标方向及触球已实现，但生成的肌肉控制未产生足以在交付片段内将球送过门线的运动，归为控制；现有证据不足以区分接触几何、脉冲参数或轨迹协调等更细子类。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志或接触对记录；视频仅 320×240，不能精确确认左右脚、首次触球身体部位、球速及骨盆高度。log step 38、46 的球坐标和速度是 agent 搜索脚本的诊断，不当作本次评分回放的逐帧状态；未据此添加跌倒约束判定。评分器未报告 time_cost，故该原生指标为 null。"
+ },
+ "summary_zh": "本次轨迹完成触球并将足球送向球门，但视频 13.0–14.0 s 的足球始终仍在门线前，踢球入门节点未达成，最终为 0/1。断点在射门操作中，现象为位姿不准，日志中的固定肌肉脉冲轨迹与视频支持控制层面的失败；缺少状态和接触日志，无法进一步确认具体肌肉参数、触球部位或骨盆高度。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "单节点 ball_in_goal 未达成（0/1）、断点在踢球操作、位姿不准 + 控制根因均有视频与日志（球最远约 x=47.6 m，距门线 2.4 m，agent 回放 success=False）支持；击打/支撑接触与实际一致。未发现需要修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:06:19Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--22mvhvw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--22mvhvw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "602462b58399f7f34cfc007cb2653db2ac0b8a879e62e81e20a3bbffec21361b",
+   "video_sha256": "0e4edd08cffc8cf9a1c13944a1c027cddb564056f48417a6b4ab24063be8456d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--d23dbdj.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--d23dbdj.json
new file mode 100644
index 0000000000000000000000000000000000000000..5b507d78dc67be668fad5b7730d66215e8c12dba
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--d23dbdj.json
@@ -0,0 +1,188 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-privileged--d23dbdj",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 1001,
+   "sim_steps": 1001,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 110,
+   "agent_resets": 14134,
+   "agent_sim_steps": 12635,
+   "agent_records": 3,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.643,
+   "peak_processes": 19,
+   "peak_disk_gb": 0.841,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-privileged--d23dbdj/replay.mp4",
+  "duration_s": 40.08,
+  "fps": 25.0,
+  "frames": 1002,
+  "sha256": "65919e670c42c8a8348051eff47c27a599ac10bbaa33c02708b90038cf3a7a36"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s，白色足球位于画面右下方起球处，远处左上方是白色球门，球尚未越过门线。"
+     },
+     {
+      "t": 2.0,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 1.60–2.60 s，模型在画面右缘降低身体、伸腿，脚部约在视频 1.9–2.1 s 与球接触，随后白球离脚向球门方向移动；视频 3.0–12.5 s，球向前滚动并逐渐减速，始终位于门线前。"
+     },
+     {
+      "t": 40.04,
+      "log_step": 127,
+      "text_zh": "观察：视频 13.0–39.0 s 的逐秒画面及视频 39.08–40.04 s 的末秒画面中，球停在球门前方草地，未到两门柱之间的门线；模型留在起球区附近，未再次触球。log step 127 中最终交付文件的生成结果为 ball_pos=[45.644721,1.298941,0.116961]，两次新进程回放均 success=false、steps=1001 且终态散列相同；评分器字段 replay_success=0、deterministic=1，与画面未进球一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频 1.9–2.1 s，模型伸出的脚部击中足球，球随即离脚前行；左右脚身份受画面右缘裁切影响，不能可靠区分。候选接触名为“击打”，现有接触方式没有踢击条目，按 S16 使用“其他”。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 1.6–2.4 s，模型另一条向画面右下方伸展的腿脚与草地接触，在低位伸腿踢球期间支承身体；全身部分被画面右缘裁切，不能可靠细分左右支撑脚。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better",
+   "value": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 2.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 2.0 s，模型伸脚踢出足球，球随后滚向球门；视频 12.0–13.0 s，球已明显减速并停在门线前，视频 39.08–40.04 s 仍在原处，未有第二次触球补救。推断：该次射门没有把球送至门线，未修复的失败步骤是“踢球入门”，以踢球时刻作为断点导航锚。"
+  },
+  "unmet": {
+   "ball_in_goal": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 12.0,
+    "log_step": 127,
+    "evidence_zh": "观察：视频 2.0–12.5 s，模型踢出的球逐渐减速，最终落点及停止位置在球门前，未越过两柱之间的门线；视频 40.04 s 仍未进门。log step 127 的最终交付生成结果报告球心 x=45.644721 米，而任务 DAG 的门线为 x=50 米，差约 4.36 米，与画面中的短射相符。推断：可见的射门结果偏差是射程不足。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "t": 2.0,
+    "log_step": 127,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "观察：log step 118 的生成脚本把优化得到的肌肉指令按控制步组成动作表并录制保存；log step 126 重新生成最终文件并进行两次新进程回放，log step 127 两次均未进球且终态散列相同，最终球心 x=45.644721 米。视频约 2.0 s 确认模型实际触球，之后球在门线前停下；agent 在 log step 123 也自述“球停在距离门线约 4.36 米处”，没有声称完成。推断：交付的肌肉控制序列能实现接触，但未产生足以完成进球的射门运动；现有证据只能定位到控制层，不能确定具体肌肉、接触几何或反馈环节。"
+   }
+  ],
+  "gap_zh": "观察：inputs/ 中没有逐帧状态日志；视频分辨率为 320×240，模型部分位于画面右缘，无法可靠区分左右脚、测出接触力或骨盆高度。推断：现有视频和最终交付回放足以确认已触球而未进球，但不足以细分控制根因，也不据低伏外观判定骨盆高度阈值被触发。评分器 reward 未报告 time_cost，原生指标值保留 null；视频时长不作为模拟进球用时。"
+ },
+ "summary_zh": "观察：模型在视频约 2.0 s 踢中足球，但球逐渐减速并停在门线前，视频 40.04 s 仍未进门，“踢球入门”节点最终为 0/1；log step 127 的最终交付回放两次均失败且终态一致，球心 x 约为 45.64 米。推断：断点在踢球入门的操作中，射程不足归为位姿不准，根因定位为控制序列未产生足够的射门运动；缺少逐帧状态、接触力和完整身体视图，不能进一步细分控制问题。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频 1.9–2.1 s 踢中球，球减速停在门线前直到 40.04 s，ball_in_goal=false、final=0、断点、接触、现象与根因（控制，子类不明停在“控制”）均有 log step 127（ball_pos x=45.64，两次回放失败）和画面支持；时钟用视频时间，无问题。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:08:16Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--d23dbdj/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--d23dbdj/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "e538d87026098b1f67971dd6f04a3575f516cd226cb563e4e024eb89d4a57171",
+   "video_sha256": "65919e670c42c8a8348051eff47c27a599ac10bbaa33c02708b90038cf3a7a36",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--gk5sahr.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--gk5sahr.json
new file mode 100644
index 0000000000000000000000000000000000000000..9bd8476f143639af8a2f5ccf1653b7171c77c010
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--gk5sahr.json
@@ -0,0 +1,189 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-privileged--gk5sahr",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 408,
+   "sim_steps": 408,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 98,
+   "agent_resets": 3209,
+   "agent_sim_steps": 337616,
+   "agent_records": 0,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-privileged--gk5sahr/replay.mp4",
+  "duration_s": 16.36,
+  "fps": 25.0,
+  "frames": 409,
+  "sha256": "3ac3e7a5b503c437fa618a0591847888f73bf809e8c9fbc516922054f91db632"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.2,
+      "log_step": null,
+      "text_zh": "观察：视频 2.8–3.4 s 的连续帧中，MyoLegs模型从画面右侧伸出的脚扫中白色足球，足球离开原来的草地点向球门方向运动；此后视频 4–15 s 球逐渐减速，始终远在门线前。"
+     },
+     {
+      "t": 16.32,
+      "log_step": 157,
+      "text_zh": "观察：视频 15.36–16.32 s（含末帧）白球停在球门前的草地，未越过门线。log step 157 最终生成的408步交付文件返回 success=False、done=True、ball_pos=[42.255533,-0.816372,0.116964]；log step 159 对同一交付文件的两次回放均失败且终态散列相同，与视频中的短距离滚球一致。"
+     },
+     {
+      "t": 16.32,
+      "log_step": null,
+      "text_zh": "观察：成功参考视频 8.4–8.76 s 可见足球到达远端球门线附近；本次视频 16.32 s 的足球仍明显位于近端草地。参考视频只用于对比完成外观，不作为本次进球证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频 2.8–3.4 s 可见MyoLegs的脚与足球接触，随后足球向球门滚出；左右脚在此视角难以辨清。候选接触名为“击打”，现有接触方式词表没有踢击项，按规则使用其他。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 3.2,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 3.2 s 模型触球后，足球只滚向前方一小段，视频 8–16.32 s 逐渐停在远离球门的位置；模型未再次接近球完成补踢。推断：失败发生在本次射门的控制与结果上，唯一目标“踢球入门”始终未修复；log step 159 的两次交付回放也均报告未进球。"
+  },
+  "unmet": {
+   "ball_in_goal": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 3.2,
+    "log_step": 157,
+    "evidence_zh": "观察：视频约 3.2 s 触球后，足球沿草地移动并减速，视频 15.36–16.32 s 已停在门线前很远处，没有到达球门；log step 157 的最终交付终点为 x=42.255533 米，而DAG门线为 x=50 米。推断：此次射门的落点及滚动终点明显过短。"
+   },
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 16.32,
+    "log_step": 159,
+    "evidence_zh": "观察：视频 3.2–4.6 s 模型可见的双腿降至近乎贴地伸展，视频 15.36–16.32 s 仍未恢复站姿，骨盆在画面右侧之外；log step 159 的最终交付回放在408个控制步返回 failure: fallen, or the time limit passed without a goal。推断：任务文本规定每控制步为仿真 0.01 s，408步仅仿真 4.08 s，尚未到仿真 10 s 时限，故本次终止对应骨盆跌倒阈值。视频共409帧、25 fps，包含初态及逐控制步画面，终止在视频 16.32 s；低位腿姿已与视频 4.0 s 附近帧核对，阈值越界本身不可直接看见。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 155,
+    "source_zh": "日志",
+    "evidence_zh": "推断：模型的肌肉驱动控制未实现足够的射门行程并维持本体支撑。log step 155 的最终生成脚本按固定步号切换三组肌肉激励，控制步100后仅激励[279,287,288,289]，不依据球速或骨盆状态调整动作；log step 157 重新生成的交付只使球到达 x≈42.26 米，log step 159 确认相同文件两次确定性失败，与视频约 3.2 s 触球后短距离滚球及双腿持续低位的结果一致。证据支持肌肉控制未奏效，尚不能证明某个具体参数、接触布局或反馈算法是唯一原因。"
+   }
+  ],
+  "gap_zh": "无逐帧状态日志；视频的模型骨盆和部分躯干在画面外，无法直接读取骨盆高度、确定失衡起点或辨清触球脚的左右及同时支撑脚的承重情况。跌倒终止由最终交付回放步数、任务仿真步长和终止条件联合推断；time_cost未由评分器报告，保留null。控制子类型缺少可验证的因果证据。"
+ },
+ "summary_zh": "本次交付在视频约 3.2 s 用脚触及足球，但球仅短距离滚动后停在门线前，“踢球入门”未达成，最终节点为0/1。固定分阶段的肌肉控制未完成有效射门及本体支撑，最终交付的两次回放均确定性失败；由408个控制步仅为仿真 4.08 s 推断终止来自跌倒。无状态日志且骨盆在画面外，具体失衡起点及控制子原因不能精确确认。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "ball_in_goal 未达成（末帧球仍在近端草地，log step 157/159 success=False，球 x≈42.26<50）、时间 3.2 s 触球、contact_actual（腿/躯干·其他）、断点和两个现象（位姿不准、平衡）均有视频与 log 支持，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:09:47Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--gk5sahr/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--gk5sahr/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ad61c33e517440bcc6e251d5a42483638158ac0d75c2980daba9f4ec6913e362",
+   "video_sha256": "3ac3e7a5b503c437fa618a0591847888f73bf809e8c9fbc516922054f91db632",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--hakzv5l.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--hakzv5l.json
new file mode 100644
index 0000000000000000000000000000000000000000..a371829c1a188b55ecc4e437a2e328902ca18ac0
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--hakzv5l.json
@@ -0,0 +1,169 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-privileged--hakzv5l",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 219,
+   "sim_steps": 219,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 201,
+   "agent_resets": 4717,
+   "agent_sim_steps": 48558,
+   "agent_records": 4399,
+   "agent_replays": 9,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.903,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.262,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-privileged--hakzv5l/replay.mp4",
+  "duration_s": 8.8,
+  "fps": 25.0,
+  "frames": 220,
+  "sha256": "7dd9605ee56a73e2ea33eb7514c4678adfb63eeae7846e4c06d08901269ab0cc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [
+     [
+      8.76,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s，黑白足球位于画面右下方起球处，远离左上方球门，进球条件初始不成立。"
+     },
+     {
+      "t": 3.6,
+      "log_step": null,
+      "text_zh": "观察：视频 3.20–3.92 s 的逐帧及脚球局部图中，模型伸出的脚靠近足球，视频约 3.60 s 脚与球相接，随后足球离开脚部向左上方球门运动；另一只脚留在草地上支承。"
+     },
+     {
+      "t": 8.76,
+      "log_step": 304,
+      "text_zh": "观察：视频 7.80–8.76 s 逐帧显示足球继续向球门滚动，从守门员画面右侧接近门线，视频 8.76 s 末帧球位于两柱之间、横梁下方的门线处；log step 303 的生成脚本保存了219步交付轨迹，log step 304 官方回放器对该文件两次均返回 success=true、steps=219、相同终态散列，reward 的 success、replay_success、deterministic、instance_ok 均为1。"
+     },
+     {
+      "t": 8.76,
+      "log_step": 304,
+      "text_zh": "推断：结合末段足球到达门线的画面和交付文件的官方成功回放，将首次进球定位在视频末帧约 8.76 s，事件随后保留；低分辨率斜视画面不能独立测出球心过线的精确瞬间，时间为近似锚点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频约 3.60 s，模型伸出的脚击中黑白足球，使其离脚向球门运动；候选接触名“击打”，词表没有踢击对应项，故用其他。画面不足以独立确认该脚的左右侧。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 3.20–3.68 s，踢球脚前伸时，模型另一只脚仍贴在草地上，承担踢球时的地面支撑；主体在画面右缘部分出界，不由此推定骨盆高度。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "模型在视频约 3.60 s 用脚击球，足球随后滚向球门，唯一节点“踢球入门”在视频末段约 8.76 s 达成并保留，最终完成1/1；交付文件在 log step 304 的两次官方回放均成功且终态一致。视频分辨率较低，模型主体部分出界，又无状态日志，球心过线精确时刻和骨盆高度不能独立量测，进球终值结合评分器确认；time_cost 未在 reward 中报告，保留为 null，视频秒数不当作仿真用时。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "单节点进球事件、末帧约8.76 s、击球接触（其他+击打、支撑）与评分器成功结果一致；无失败，无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:08:39Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--hakzv5l/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--hakzv5l/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "af86b22806a35b60cd65cb1b2a08774dd7ba6c13376c378febb1a23d75c16920",
+   "video_sha256": "7dd9605ee56a73e2ea33eb7514c4678adfb63eeae7846e4c06d08901269ab0cc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--ixgwfzy.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--ixgwfzy.json
new file mode 100644
index 0000000000000000000000000000000000000000..9d975dae1af9c632f90abbd8782d369258ad37e8
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--ixgwfzy.json
@@ -0,0 +1,193 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-privileged--ixgwfzy",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 195,
+   "sim_steps": 195,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 93,
+   "agent_resets": 14190,
+   "agent_sim_steps": 16999,
+   "agent_records": 0,
+   "agent_replays": 8,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.746,
+   "peak_processes": 29,
+   "peak_disk_gb": 0.983,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 1,
+   "session_log_from_monitor": 1
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-privileged--ixgwfzy/replay.mp4",
+  "duration_s": 7.84,
+  "fps": 25.0,
+  "frames": 196,
+  "sha256": "e36a3ee79385133282dcd1c909fbc9c9805ee9a2c7b2ca9408000caebc8bd8a9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 足球位于画面右下方起球处，远离左上方球门，初态不满足进球条件；按视频每 0.4 s 的全程抽帧查看，球一直没有到达门线。"
+     },
+     {
+      "t": 4.8,
+      "log_step": null,
+      "text_zh": "观察：逐帧查看视频 4.48–5.40 s，模型靠近足球的下肢末端与球接近并相接，随后球向球门方向离开原位；模型同时明显前倾，球只在起球区域附近移动。"
+     },
+     {
+      "t": 7.8,
+      "log_step": 204,
+      "text_zh": "观察：视频 6.88–7.80 s 的逐帧画面中，足球仍在近处草地，未进入球门，模型逐渐倒地；log step 204 明确重新生成最终 195 行交付轨迹，脚本输出 ballmax=41.44673372753346、crossed=False，回放服务两次均返回 success=false、steps=195 且 state_sha 相同，与评分器 n_actions=195、deterministic=1 一致。"
+     },
+     {
+      "t": 7.8,
+      "log_step": null,
+      "text_zh": "观察：已对照成功参考视频全程抽帧及参考视频 8.76 s 末帧，参考球最终到达门柱处；本次视频 7.80 s 的球与门线之间仍有大片草地，缺少进球的可见结果。参考视频仅用于辨认球门位置，不作为本次进度证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频约 4.48–5.00 s，模型靠球侧的下肢末端短促触球，随后足球向前移开；候选接触名“击打”，词表无踢击对应项，使用其他。低分辨率斜侧视角不能可靠区分左右脚，不按脚本预定右脚补认身份。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频约 3.52–5.00 s，模型摆动近球侧下肢时，另一侧脚部仍落在后方草地上承重；随后躯干继续前倾下沉，至视频 7.80 s 倒在草地上。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 4.0,
+   "log_step": 204,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 4.00 s 模型在靠近足球并摆腿过程中显著前倾，视频 4.48–5.40 s 触球后也未恢复直立，视频 6.88–7.80 s 持续下沉并倒地，球始终远在门线前；log step 204 对最终 195 行轨迹的两次回放均报告 failure: fallen, or the time limit passed without a goal。此处断点是正在尝试而未完成的踢球入门步骤，视频约 4.00 s 仅为失衡过程的粗略导航锚点。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 4.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 3.52–4.44 s 模型躯干逐步前倾、腿部拖在后方；视频 4.48–6.36 s 触球后仍低伏而未重新站稳，视频 6.88–7.80 s 躯干进一步下沉，最后倒在草地上，足球尚未接近球门。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 147 的最终控制器实现采用关节 PD、重心参考前移和定时抬腿／摆腿／屈伸收尾，log step 176 的交付生成脚本调用该 rollout；log step 204 将最后候选重新生成并验证为同一条 195 行失败轨迹。结合视频约 4.00–7.80 s 前倾持续发展且始终未恢复支撑，说明该肌肉控制运动未能在触球过程中维持或恢复全身稳定；现有证据不足以进一步指定某个增益、肌肉力矩映射或反馈环节。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志，无法核实触球的具体身体链接、接触力、骨盆高度阈值触发的准确时刻，也不能把全身稳定失败细分为某项参数或反馈实现错误。time_cost 未在评分器 reward 字段报告，保持 null；视频时间与模拟时间不同，不以视频时长填充该指标。会话早期其他候选的成绩和跌倒诊断不当作最终交付的节点进度。",
+  "unmet": {
+   "ball_in_goal": false
+  }
+ },
+ "summary_zh": "观察：本次模型实际尝试了踢球，视频约 4.48–5.40 s 足球受下肢接触后仅向前移动一小段，至视频 7.80 s 仍远在门线前，踢球入门节点最终为 0/1。推断：模型在触球过程中持续前倾并最终倒地，交付运动的全身稳定控制未奏效；log step 204 的最终 195 行轨迹两次确定性回放均失败，与画面一致。无状态日志，具体触球链接、跌倒阈值时刻及更细的控制原因无法核实。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、触球接触、断点(ball_in_goal，操作中)及 平衡/控制 标注与视频一致，未发现需修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:03:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--ixgwfzy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--ixgwfzy/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "18b423d23da2f907569aade8fb8329e5584a9384b572ab1cffb3db8cf66d9ab3",
+   "video_sha256": "e36a3ee79385133282dcd1c909fbc9c9805ee9a2c7b2ca9408000caebc8bd8a9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--kwub4zn.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--kwub4zn.json
new file mode 100644
index 0000000000000000000000000000000000000000..d03861c5ab79237d02b06b02edc1407038303a3a
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--kwub4zn.json
@@ -0,0 +1,193 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-privileged--kwub4zn",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 235,
+   "sim_steps": 235,
+   "task_failure": 1,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 170,
+   "agent_resets": 142,
+   "agent_sim_steps": 15749,
+   "agent_records": 2,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.226,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.241,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-privileged--kwub4zn/replay.mp4",
+  "duration_s": 9.44,
+  "fps": 25.0,
+  "frames": 236,
+  "sha256": "cae0df2bbe066b5292aa01610271f8a50f19a3846099684d9c685b2a9725b0c1"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s，足球位于画面右下方起球处，远端球门在左上方，初态未进球。"
+     },
+     {
+      "t": 7.6,
+      "log_step": null,
+      "text_zh": "观察：视频 5.6–7.6 s，MyoLegs 模型向球前倾并逐渐下沉，足球仍在起球处附近；视频 7.0–8.4 s 的逐帧放大画面没有显示足球被踢向球门。"
+     },
+     {
+      "t": 9.4,
+      "log_step": 89,
+      "text_zh": "观察：视频 8.4–9.4 s，模型倒在足球附近，球未沿球场到达远端门线，两门柱间没有进球事件；log step 89 对最终保存的 235 行动作回放返回 success=False、solved=0.0，评分器也报告 replay_success=0、deterministic=1。"
+     },
+     {
+      "t": null,
+      "log_step": 126,
+      "text_zh": "观察：log step 89 保存最终动作文件，log step 112 仅同步生成脚本参数，log step 126 检查文件仍为 235×290 动作；log step 124–125 的后续试验及 log step 127 自述的触球改进不能作为该交付轨迹已经实现脚踢的证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 5.6–7.2 s，模型可见的脚与球场地面接触，支承向球倾身的动作；随后身体降低，视频 8.4–9.4 s 腿和躯干落至地面。"
+     },
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频 7.6–9.4 s，足球与模型身体在低分辨率斜侧视图中重叠，球仅留在原位附近；无法确定是否发生脚踢、身体挤碰或接触的具体部位与方式，不把后续会话测试中的脚踢算入本轨迹。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 7.2,
+   "log_step": 89,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 7.2 s，模型在向足球倾身、尝试踢球的过程中支撑姿态明显塌下；视频 7.6–9.4 s 身体继续下降并倒地，没有恢复，也没有足球到达球门。log step 89 回放并保存的交付动作以 failure: fallen, or the time limit passed without a goal 结束。"
+  },
+  "unmet": {
+   "ball_in_goal": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 7.2,
+    "log_step": null,
+    "evidence_zh": "观察：视频 6.4–7.6 s，模型前倾角度持续增大、身体高度降低；视频 8.4–9.4 s 躯干和双腿落在草地上，未恢复站立，足球仍远离球门。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：log step 88 的交付生成脚本用踝部位置与速度反馈推动身体前移，同时给右髋、膝、踝设置摆腿目标；log step 89 修改摆腿参数后回放保存，仍返回失败。对应视频 6.4–9.4 s 显示身体在摆腿阶段持续前倾、下沉直至倒地，说明该肌肉控制序列未能协调支撑与踢球动作；现有证据不足以进一步区分轨迹求解、反馈控制或具体参数问题。"
+   }
+  ],
+  "gap_zh": "没有逐帧状态日志或接触对；视频 7.6–9.4 s 足球与身体重叠，无法可靠分辨触球部位、接触方式和骨盆触发 0.2 米阈值的精确时刻，因此不追加未触球或击球力度方面的现象。评分器 reward 未报告 time_cost，原生指标保留 null；视频播放时间不当作仿真用时。"
+ },
+ "summary_zh": "评分轨迹未完成“踢球入门”（最终 0/1）：视频至 9.4 s 足球仍在起球处附近，未越过远端门线，与评分器失败结果一致。模型在视频约 7.2 s 的踢球尝试中持续失衡并倒地，交付控制序列未能协调支撑与摆腿。日志后续触球试验未写入最终动作文件；无状态日志且球旁接触遮挡，具体触球方式仍不确定。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、失败为平衡、断点7.2 s与视频一致；球始终留在起球处附近，模型从画面右缘倾倒。未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:05:12Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--kwub4zn/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--kwub4zn/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "32e68985dc4c3cf9842b5250ac96d6fdd58ff0da72cb6b4e3c757c1def2a680f",
+   "video_sha256": "cae0df2bbe066b5292aa01610271f8a50f19a3846099684d9c685b2a9725b0c1",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--z9xio66.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--z9xio66.json
new file mode 100644
index 0000000000000000000000000000000000000000..71913bbed0e44128ac95a51a9c7884a2d3949ed5
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-privileged--z9xio66.json
@@ -0,0 +1,169 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-privileged--z9xio66",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "n_actions": 318,
+   "sim_steps": 318,
+   "task_failure": 0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 73,
+   "agent_resets": 77,
+   "agent_sim_steps": 1635,
+   "agent_records": 0,
+   "agent_replays": 4,
+   "agent_cost_self_reported": 1,
+   "infra_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.322,
+   "peak_processes": 24,
+   "peak_disk_gb": 0.243,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-privileged--z9xio66/replay.mp4",
+  "duration_s": 12.76,
+  "fps": 25.0,
+  "frames": 319,
+  "sha256": "aeb98f68268e3ef42619bdb4b5b77dc8e600bc9a244a15f595f837984009b64b"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [
+     [
+      12.72,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s，白色足球位于画面右下方的起球处，远离左上方球门，进球条件初始不成立。"
+     },
+     {
+      "t": 5.6,
+      "log_step": null,
+      "text_zh": "观察：视频 4.48–5.60 s 的逐帧局部图中，MyoLegs 的脚部接近并接触足球；视频约 5.4–5.6 s，球开始明显离开脚部向球门滚动。视频 6.0–11.5 s，球持续沿草地靠近球门，模型留在起球处附近、下肢折叠且身体降低。"
+     },
+     {
+      "t": 12.72,
+      "log_step": 82,
+      "text_zh": "观察：视频 11.72–12.72 s 的逐帧球门局部图显示足球沿地面到达画面近侧门柱内的门线处，低于横梁且在守门员旁；末帧球心位于门线附近的球门侧。log step 82 中，基准回放工具对已保存的交付文件重复回放两次，均报告 success=true、steps=318、instance_ok=true、deterministic=true，终态散列相同；run.json 的评分字段也报告 success=1、replay_success=1。"
+     },
+     {
+      "t": 12.72,
+      "log_step": 82,
+      "text_zh": "推断：结合球到达门线的末帧与评分器的进球成功结果，将事件首次完成锚定为视频约 12.72 s，并按事件节点保持至结尾。交付回放的 318 个控制步与含初态的 319 帧、25 fps 对应，第 318 步映射为视频 318/25=12.72 s，已在视频末尾足球到达门线处核对；不是将 agent 工作时间或模拟秒数直接换成视频秒数。原始视频仅 320×240 且没有状态日志，球心精确坐标与越线控制步依赖评分器确认。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频约 5.4–5.6 s，MyoLegs 的脚部击打足球并使其离开起球处；局部遮挡和分辨率不足以独立分清左右脚。候选接触名为“击打”：词表没有踢击方式，故使用其他。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 4.48–5.60 s，模型踢球时另一下肢接地；踢出后下肢屈曲贴近草地，身体降低并继续由下肢与地面接触承托（视频 6.0–12.72 s）。具体左右脚及各段承重比例无法从该视角分辨。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "本次交付完成“踢球入门”事件节点，最终为 1/1：视频约 5.4–5.6 s，MyoLegs 用脚击出足球，球沿地面滚向球门，在视频约 12.72 s 到达近侧门柱内的门线处。评分字段与 log step 82 的两次确定性回放均确认进球成功，无失败断点。没有状态日志，低分辨率画面不能独立测量球心越线坐标或精确区分触球脚的左右；完成锚点结合末帧与评分器确认，time_cost 未在评分字段中报告，保留为 null，视频时长不等于模拟用时。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，单一事件节点 ball_in_goal 在视频约 12.72 s 到达，与评分字段（success=1，318 步，回放确定性）及末帧球位于近侧门柱旁门线处一致；踢球时间约 5.4–5.6 s 与帧相符，接触标注合理，无失败断点。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:04:15Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--z9xio66/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-privileged--z9xio66/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "140c650988f8a81ae2694236579cf191c6ed2b9fe087ed5df91c61d359127d26",
+   "video_sha256": "aeb98f68268e3ef42619bdb4b5b77dc8e600bc9a244a15f595f837984009b64b",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--5bmdbus.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--5bmdbus.json
new file mode 100644
index 0000000000000000000000000000000000000000..c4b0b0d9dd51f97fd5dfcd6b6dc6a57ef8c5d5ca
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--5bmdbus.json
@@ -0,0 +1,180 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-standard--5bmdbus",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "physics_steps": 0,
+   "steps": 0,
+   "calls": 1,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 0,
+   "elapsed_s": 3644.0,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.55,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.389,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-standard--5bmdbus/replay.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "c095c8c4402e07b650b6e43d275fe7fe3aabb9ac1bc2b4b636a24bb6cec051bb"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：本次视频仅有视频 0.00 s 的一个初始帧，足球仍在画面右下方起球区域，远离左上方球门的门线，没有进球。已检查全部一帧；参考成功视频在其视频 8.76 s 显示足球到达门线处，仅用于确认视角中的完成位置。"
+     },
+     {
+      "t": 0.0,
+      "log_step": 5,
+      "text_zh": "观察：log step 5 中在线机器人服务的初始状态为 episode=1、control_steps=0；评分器最终记录 n_actions=0、sim_steps=0、resets=0、live_success=0、replay_success=0、live_matches_replay=1，受评分回合始终没有推进。"
+     },
+     {
+      "t": null,
+      "log_step": 88,
+      "text_zh": "观察：log step 88 的踢球、骨盆降低和足球过线数值来自 agent 自建的本地 MuJoCo 调试脚本，而非在线受评分回合；这些会话测试不计为本节点的进球或实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 0.0,
+   "log_step": 5,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0.00 s 为未踢球的初始场景；log step 5 在线服务读取到 control_steps=0，评分器最终 n_actions=0、sim_steps=0、episodes=1、resets=0。该在线初始回合被验收为失败，踢球入门步骤未启动；log step 88–91 的动作仅在本地调试中。"
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 5,
+    "evidence_zh": "观察：视频 0.00 s 的唯一帧保留起球初态，评分器 n_actions=0、physics_steps=0；log step 5 只获取在线服务规格、状态和初始观测，后续日志没有提交在线动作，受评分回合没有开始踢球。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "观察：本次视频存在但仅有初始一帧，无状态日志；评分器未报告 time_cost。log step 88 的本地候选控制在本地仿真 1.18 s 骨盆过低，足球于本地仿真 2.37 s 才过线；log step 90–91 仍在启动并等待本地恢复动作搜索，日志没有其最终结果，也没有说明最终为何未向在线回合提交动作。推断：这些会话证据不足以确定未启动受评分步骤的具体根因，因此根因层留空；不把本地跌倒、预算耗尽或缺少动作提交本身当作根因。",
+  "unmet": {
+   "ball_in_goal": false
+  }
+ },
+ "summary_zh": "受评分在线回合始终停留在初态，踢球入门节点未达到，最终进度为0/1；唯一视频帧与评分器记录的0次动作、0个仿真步一致。断点为踢球入门的首步前，现象为没尝试；日志中的本地踢球测试没有进入受评分回合。日志结束于本地恢复动作搜索的等待报告，无法确定未提交在线动作的具体根因，故根因留空。",
+ "conf": 9,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标注与证据一致：视频仅一帧（0.00 s），球在右下起球处未进门；评分器 n_actions=0、sim_steps=0，受评分回合未启动，节点未达、断点首步前、现象没尝试符合 CASES FAIL-22 / OPEN-16（agent 自己停下且受评分回合零步），根因留空并写 gap_zh。本地调试脚本未被当作事实，时钟无换算。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:09:53Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--5bmdbus/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--5bmdbus/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6166d92652b375abb350c5374ed97236c8bea477440a95b2f3905040e52454ef",
+   "video_sha256": "c095c8c4402e07b650b6e43d275fe7fe3aabb9ac1bc2b4b636a24bb6cec051bb",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--5er2pbw.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--5er2pbw.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec872e4e598189ec3f20660bf9305426731595db
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--5er2pbw.json
@@ -0,0 +1,207 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-standard--5er2pbw",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 1000,
+   "sim_steps": 1000,
+   "physics_steps": 10000,
+   "steps": 300,
+   "calls": 19,
+   "episodes": 1,
+   "resets": 0,
+   "reset_budget": 10,
+   "sim_steps_total": 1000,
+   "elapsed_s": 3294.8,
+   "video_rendered": 1,
+   "task_failure": 0,
+   "timeout": 1,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.826,
+   "peak_processes": 13,
+   "peak_disk_gb": 0.383,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-standard--5er2pbw/replay.mp4",
+  "duration_s": 40.04,
+  "fps": 25.0,
+  "frames": 1001,
+  "sha256": "391c7dd474022222c890dab450668b37e63e45ccdd119b672d4347bfadc6e40f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.6,
+      "log_step": 342,
+      "text_zh": "观察：视频 3.28–4.16 s，MyoLegs 模型在起球处侧倒，前伸的腿部与足球接近并重叠，随后足球离开原位向左上方球门方向滚动；这是本次在线执行的踢球尝试。log step 342 的服务返回显示已执行300步肌肉控制并观察到累计700控制步，回合未重置。"
+     },
+     {
+      "t": 14.0,
+      "log_step": null,
+      "text_zh": "观察：视频 6.0、8.0、10.0、12.0、14.0 s，足球逐渐减速，停在球门前的场地上；视频 16.0–38.0 s 的每隔2秒画面及视频 39.04–40.00 s 的末秒画面中，足球始终在门线前，模型保持侧倒姿势，没有再次击球。"
+     },
+     {
+      "t": 40.0,
+      "log_step": 348,
+      "text_zh": "观察：视频 40.00 s，白球与门线仍有明显距离；评分器字段 live_success=0、success=0，log step 348 的服务状态为 control_steps=1000、episode=1、control_steps_total=1000。进球事件未发生；并非只由零分推断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频 3.52–4.24 s，模型前伸腿部碰到足球，足球随后向球门滚出。候选接触名为“击打”，词表无踢击对应项，故用其他；远景分辨率不足以精确区分小腿或脚的接触面，不采用会话本地测试的接触对作为本回合事实。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 3.5–5.0 s，模型由站立运动转为侧倒，腿和躯干落到场地上承重，此后直至视频 40.00 s 都没有恢复站立。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 3.6,
+   "log_step": 342,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 3.5–4.3 s，模型在踢球入门尝试中侧倒，腿部与球接触后球向前滚出；视频 14.0 s 及视频 39.04–40.00 s 显示球停在门线前，模型一直未起身补踢。该步骤始终未修复，评分器 success=0、timeout=1、task_failure=0。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 3.6,
+    "log_step": 343,
+    "evidence_zh": "观察：视频 3.28–4.5 s，模型骨盆和躯干降至地面附近并侧倒，视频 5.0–40.00 s 一直躺在起球处；log step 343 的在线服务本体读数 root_z=0.2069米。这里标可见的失衡倒地，不把它说成评分器已触发骨盆低于0.2米的终止条件。"
+   },
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 14.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 4.0–12.0 s，足球离开腿部后向球门方向滚动并减速；视频 14.0–40.00 s，球停在门线前，未到达两柱间的门线区域，踢球的最终位置偏短。球的前进路线上未见守门员或门柱阻挡。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：log step 338 的已交付脚本按固定时间生成300行肌肉指令，只有起蹬和后续持续激活两段，没有根据身体姿态或球的位置修正动作；log step 342 执行该脚本后仅继续观察，log step 347 再观察到1000控制步。结合视频 3.5–4.5 s 的侧倒及视频 14.0–40.00 s 的球停在门前，说明这套肌肉控制未协调好身体支撑与有效踢球，也未恢复后补踢。证据不足以把问题进一步归到特定肌肉参数、接触几何或反馈跟踪子类。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "本次没有逐帧状态日志，远距离单视角且模型初始大部分位于画面右边缘，不能精确确定脚或小腿的触球面、接触力和失衡起始时刻；断点取视频约 3.6 s 的可见倒地与触球过程。log step 329 的胫骨接触记录及 log step 315 的球坐标来自agent本地模拟测试，不是评分回合，未用于本次节点或实际接触判定。评分器未报告 time_cost，保留null；不能将视频40.04秒当作仿真40.04秒。",
+  "unmet": {
+   "ball_in_goal": false
+  }
+ },
+ "summary_zh": "本次在线回合尝试踢球，模型在视频约 3.5–4.5 s 侧倒并以腿部击出足球；球向球门滚动后停在门线前，至视频 40.00 s 仍未进球，最终完成0/1节点。断点是“踢球入门”的操作过程，现象为平衡丧失和球的位置偏短，固定肌肉控制没有完成支撑、踢球及恢复协调。无逐帧状态日志且远景遮裁了部分本体，无法细分触球面或肌肉控制根因；不将本地模拟测试当作评分回合。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点 ball_in_goal 未达成、断点（踢球操作中，约3.6 s 侧倒触球）、接触（腿/躯干 其他+支撑）与视频一致；球停在门线前，回合至1000步（视频40 s）未终止、未进球。未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:08:13Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--5er2pbw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--5er2pbw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f963690aa5d9e783eb4502d2d1035d99a2a0f47ffeb4b5deefbc2abcce8b8798",
+   "video_sha256": "391c7dd474022222c890dab450668b37e63e45ccdd119b672d4347bfadc6e40f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--8m3ozuq.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--8m3ozuq.json
new file mode 100644
index 0000000000000000000000000000000000000000..aeff298fe89d93163b5d984c24c385d1da05e3a1
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--8m3ozuq.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-standard--8m3ozuq",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 163,
+   "sim_steps": 163,
+   "physics_steps": 1630,
+   "steps": 1508,
+   "calls": 473,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 1508,
+   "elapsed_s": 3798.9,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-standard--8m3ozuq/replay.mp4",
+  "duration_s": 6.56,
+  "fps": 25.0,
+  "frames": 164,
+  "sha256": "56d05edc56dad2a1cb2d5fa9dff0371d7070722a876e595c16981a9c1df14719"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.00 s 的足球位于画面右下方起球处，远端球门在左上方，进球条件初始不成立。"
+     },
+     {
+      "t": 5.4,
+      "log_step": null,
+      "text_zh": "观察：视频 3.20–5.40 s，MyoLegs 模型向足球靠近、弯躯并转动双腿，球仍在起球处附近，未见足球沿球场向远端球门运动；视频 5.40–6.52 s 模型继续翻倒，未见球进入门线区域。"
+     },
+     {
+      "t": 6.52,
+      "log_step": 141,
+      "text_zh": "观察：视频 6.52 s 末帧模型已翻倒；log step 141 的服务输出为 episode 11、control step 163、done=True、root 高度 0.198米，原因字段为“failure: fallen, or the time limit passed without a goal”。log step 142 确认最后回合终止且无重置余量；评分字段 live_success=0、replay_success=0、live_matches_replay=1。"
+     },
+     {
+      "t": 6.52,
+      "log_step": 142,
+      "text_zh": "推断：log step 142 记录最后回合共163个100Hz控制步，仿真用时1.63 s；本视频共164帧（含初态），控制步按25帧每秒播放，末步对应视频 163/25=6.52 s，已与视频末帧翻倒及回合结束核对。结合全片未见足球向球门推进，节点从未达成；不是依据零分单独推定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 3.20–5.40 s，MyoLegs 双脚在球场地面交替落地、承担靠近足球和摆腿时的身体支撑；随后模型翻倒。该视角无法可靠区分承重脚的左右，末段身体遮住球附近，未确认可单独记入的脚球击打接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 5.4,
+   "log_step": 138,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 5.40 s，MyoLegs 在足球旁摆腿时身体明显倾斜并继续翻转，视频 5.92–6.52 s 已倒向地面，足球没有被送入远端球门。log step 138、140 的最后回合本体读数显示 root 高度依次为0.705米、0.498米，log step 141 降至0.198米并结束回合；踢球入门这一步未获修复。"
+  },
+  "unmet": {
+   "ball_in_goal": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 5.4,
+    "log_step": 141,
+    "evidence_zh": "观察：视频 5.00–6.52 s，MyoLegs 在球旁前倾、旋转，双腿未能维持身体支撑，最后躯干近乎水平翻倒；log step 141 的服务本体读数为 root 高度0.198米并报告回合失败。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 124,
+    "source_zh": "日志",
+    "evidence_zh": "推断：log step 124 的最后回合控制器根据骨盆姿态、角速度及高度误差计算支撑腿力矩并分配肌肉激励，目标高度为0.86米；log step 127、131、137、139 又修改支撑补偿和肌肉激励。可是 log step 132、138、140、141 的本体反馈显示高度由0.815米持续降到0.705米、0.498米、0.198米，视频 5.00–6.52 s 同时可见倾倒，说明实际肌肉控制未能在摆腿时维持支撑。证据不足以将其进一步归到某一参数、逆解或接触几何。"
+   }
+  ],
+  "gap_zh": "本目录没有状态日志。视频分辨率低，视频约 5.40–6.52 s 球附近被模型身体遮挡，无法确定是否发生短暂脚球或躯干球接触，因此只记录可确认的地面支撑，不追加击打或没建立接触现象。日志可支持控制层归因，但不能定位导致失稳的具体肌肉或参数；评分器未报告 time_cost，故该指标为null。"
+ },
+ "summary_zh": "最后评分回合中，MyoLegs 靠近足球并摆腿，但球未进入远端球门，“踢球入门”节点最终为0/1。视频约 5.40 s 起可见明显失衡并持续翻倒，最后回合日志确认骨盆高度降至0.198米后失败；控制器的支撑与姿态修正未能维持平衡。末段身体遮挡球附近且无状态日志，不能确认短暂触球或进一步细分控制根因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 8,
+    "summary_zh": "节点未达成、断点、平衡现象和控制根因与视频及日志一致；仅有 log 中 root 高度读数引用错误，需更正数字。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点、平衡现象与控制根因均与视频和日志一致；球全程留在起球处附近，模型 5.4 s 后翻倒，step 141 终止读数 root 高度0.198米与视频相符。上一轮的数值争议已由标注方以日志字段解释，无新证据。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:13:20Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--8m3ozuq/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--8m3ozuq/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--8m3ozuq/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--8m3ozuq/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f563abf832878b04b496849d2aa455c7325a0933efbce7ce4b5f5644c9ea4d6c",
+   "video_sha256": "56d05edc56dad2a1cb2d5fa9dff0371d7070722a876e595c16981a9c1df14719",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--cs9f7ib.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--cs9f7ib.json
new file mode 100644
index 0000000000000000000000000000000000000000..f4bc0ba8509b4e37b9694bad023467ac9ff79890
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--cs9f7ib.json
@@ -0,0 +1,200 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-standard--cs9f7ib",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 125,
+   "sim_steps": 125,
+   "physics_steps": 1250,
+   "steps": 529,
+   "calls": 8,
+   "episodes": 4,
+   "resets": 3,
+   "reset_budget": 10,
+   "sim_steps_total": 529,
+   "elapsed_s": 3107.0,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "grader_error": 0,
+   "instance_mismatch": 0,
+   "infra_error": 0,
+   "gave_up": 0,
+   "sim_error": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 2.173,
+   "peak_processes": 28,
+   "peak_disk_gb": 1.135,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-standard--cs9f7ib/replay.mp4",
+  "duration_s": 5.04,
+  "fps": 25.0,
+  "frames": 126,
+  "sha256": "5e1a516ca5c4bc21aa4e0948fa3062b6ac051ec3147874dd3ba6404d0b10df97"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 白色足球在画面右下方起球处，球门在远端左上方；视频 0.0–5.0 s 的全程抽帧及视频 2.8–5.0 s 的逐帧检查中，足球未到达门线。"
+     },
+     {
+      "t": 3.92,
+      "log_step": null,
+      "text_zh": "观察：视频约 3.8–4.0 s，MyoLegs 模型伸向足球的脚与球接触，随后足球离开原位向球门方向移动；视频 4.2–5.0 s，模型继续下沉，球仅离开起球处一小段距离，末帧仍远在门线前。"
+     },
+     {
+      "t": 5.0,
+      "log_step": 220,
+      "text_zh": "观察：log step 220 中 agent 的脚本重置并执行最终计划，机器人服务（记录在线回合状态的接口）返回 episode=4、control_steps=125、done=true，结束原因为“failure: fallen, or the time limit passed without a goal”；log step 223 核对时仍为同一回合。评分器字段 success=0、live_matches_replay=1 与视频未进球一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频约 3.8–4.0 s，模型前伸的脚击中白色足球，球随即离开原位；左右脚身份无法可靠区分。候选接触名为“击打”，现有接触方式词表没有踢击对应项，因此用其他。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 2.8–4.0 s，模型一脚向前伸向足球，另一条屈曲的腿以脚接触草地承重；视频 4.2–5.0 s 身体继续下沉，未恢复站立支撑。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 4.0,
+   "log_step": 220,
+   "where": "操作中",
+   "evidence_zh": "推断：视频约 3.8–4.0 s，MyoLegs 模型伸脚触球，身体随后持续向地面塌落；视频 4.2–5.0 s 未恢复支撑，球仍远在门线前。log step 220 的最终在线第 4 回合在 125 个控制步后失败结束，log step 223 确认该回合未再重置。因此未修复的失败步骤是正在执行的“踢球入门”。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 4.0,
+    "log_step": 220,
+    "evidence_zh": "观察：视频 3.4–4.0 s 模型屈腿、伸脚触球，视频 4.2–5.0 s 身体继续降低并向地面倒伏，末帧未站起；足球尚未接近球门。log step 220 中机器人服务返回该最终回合已失败结束，未出现进球成功。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 220,
+    "source_zh": "日志 / 视频",
+    "evidence_zh": "推断：agent 在 log step 165 编写按预先生成的肌肉命令逐行播放的执行脚本，log step 220 将最终优化计划实际送入第 4 回合；该回合视频约 3.8–5.0 s 显示模型虽能触球，却无法在踢球动作中维持身体支撑并继续等待球抵达门线。log step 220 的本地同计划测试也报告 fall_at 和 goal_at=None，作为会话证据支持这一控制能力缺口；不能据此细分为反馈、轨迹求解或特定增益问题。"
+   }
+  ],
+  "gap_zh": "无状态日志；录像仅为低分辨率球门斜侧视角，模型初段部分在画外，无法可靠识别踢球脚左右、精确骨盆高度及触发跌倒阈值的时刻。time_cost 未在评分器字段报告，保留 null；log step 220 的本地测试数值属于会话仿真，未当作本次视频的逐帧状态或时间。缺少接触力与跟踪误差记录，根因只标控制，不能确定子类型。",
+  "unmet": {
+   "ball_in_goal": false
+  }
+ },
+ "summary_zh": "正式计分的是 log step 220 执行的第 4 回合：模型在视频约 3.8–4.0 s 用脚触球，但球仅向球门方向移动一小段距离，唯一目标“踢球入门”始终未达成，终态为 0/1。模型在踢球过程中失去平衡并倒伏，控制未能兼顾触球与身体支撑，在线服务和评分器均判失败。无状态日志且模型部分在画外，无法确定精确跌倒阈值时刻或更细的控制根因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点 ball_in_goal 未达成（球在 5.0 s 仍在空中远离门线）、断点、平衡现象与控制根因均有视频和 log step 220/223 的服务读数支持；触球时间约 4.0 s 在一秒误差内。无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:08:42Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--cs9f7ib/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--cs9f7ib/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ca9c1b8e6fc087726913c1b98f242630c7fb78dab494f0936a0bb6c12817911f",
+   "video_sha256": "5e1a516ca5c4bc21aa4e0948fa3062b6ac051ec3147874dd3ba6404d0b10df97",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--mqhkayq.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--mqhkayq.json
new file mode 100644
index 0000000000000000000000000000000000000000..c9a34f2faaab9e55db35e06d94ca05d398fabbeb
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--mqhkayq.json
@@ -0,0 +1,195 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-standard--mqhkayq",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 85,
+   "sim_steps": 85,
+   "physics_steps": 850,
+   "steps": 752,
+   "calls": 51,
+   "episodes": 11,
+   "resets": 10,
+   "reset_budget": 10,
+   "sim_steps_total": 752,
+   "elapsed_s": 880.1,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.584,
+   "peak_processes": 12,
+   "peak_disk_gb": 0.856,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-standard--mqhkayq/replay.mp4",
+  "duration_s": 3.44,
+  "fps": 25.0,
+  "frames": 86,
+  "sha256": "3845f202828bee18033a0c62a8bceb79b014d26593504faeb58cf8f08757c5d5"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s，足球位于画面右下方起球处，远端白色球门内没有该球，进球条件初始不成立。"
+     },
+     {
+      "t": 1.12,
+      "log_step": null,
+      "text_zh": "观察：逐帧检查视频 0.0–3.40 s；视频约 0.8–1.6 s，一条腿从右侧向足球附近摆动，但可见脚端仍在球的右侧，足球未被踢离起点；视频 2.40–3.40 s 的末段也未见球向球门运动或过线。"
+     },
+     {
+      "t": 3.4,
+      "log_step": 59,
+      "text_zh": "观察：log step 59 的调用记录含最后一次 reset 与控制脚本执行；评分器记录 episodes=11、resets=10、n_actions=85、graded_by_live=1、success=0、task_failure=1。视频显示的是最后回合，前面重置掉的测试不计入节点进度。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0.0–2.4 s 右下边缘可见一只脚位于草地上支撑，另一条腿摆向球旁；无法可靠辨认左右脚。未见脚端与足球相交或把球击出，不将计划中的击打记为实际接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 1.12,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 0.8–1.6 s，模型的摆动脚在足球右侧经过，没有将球击向球门；视频直至 3.40 s 足球仍留在原处。这次踢球未命中的结果在后续画面中未修复，失败步骤为踢球入门。"
+  },
+  "unmet": {
+   "ball_in_goal": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 1.12,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.96–1.60 s，摆动脚端在足球右侧伸出并回摆，始终未与可见球体重合；逐帧检查至视频 3.40 s，球仍在起球位置。脚的摆动没有对准这个静止目标。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "source_zh": "视频",
+    "evidence_zh": "推断：视频约 0.8–2.8 s 可见模型执行摆腿，但脚端运动没有覆盖足球所在位置，球始终未被击出；结合 log step 58 中 agent 明确提出低位随摆踢球、log step 59 执行最后回合控制脚本，支持踢球动作执行不当。现有记录不足以区分肌肉指令参数、轨迹生成和反馈稳定性的具体问题。"
+   }
+  ],
+  "gap_zh": "没有状态日志；录像只有低分辨率球门斜侧视角，模型躯干和骨盆处于右侧画外，左右脚身份及精确接触力不可核实。agent 自述“最终回合因骨盆跌落而失败”（log step 62），但对应服务状态输出在 log step 61 为空，不能据此把画外跌倒另标为已观察到的平衡现象。最后控制脚本正文未完整保留，无法进一步归因；评分器未报告 time_cost，故其值为 null。"
+ },
+ "summary_zh": "评分回合未完成踢球入门，最终进度为 0/1：视频 0.0–3.40 s 足球始终留在起球处。视频约 1.12 s 的摆脚未对准足球，断点为踢球入门的操作中，根因可支持到控制层。躯干和骨盆在画外，无法独立核实 agent 在 log step 62 所称的跌倒，具体控制失效机制也缺少记录。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频全程球留在起球处，ball_in_goal 未达成、进度 0/1 正确；断点、位姿不准（摆脚在球右侧掠过）和停在控制层的根因均有帧证据支持，且正确说明了画外躯干与 step 62 自述不可核实。无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:13:38Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--mqhkayq/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--mqhkayq/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a43039f4615de8b3ed97518989e9fdfe8b4f8ab882b34cffb6296b055a6767d1",
+   "video_sha256": "3845f202828bee18033a0c62a8bceb79b014d26593504faeb58cf8f08757c5d5",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}
diff --git a/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--sys8e3i.json b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--sys8e3i.json
new file mode 100644
index 0000000000000000000000000000000000000000..ee556f24ba621eda2edb603a2bdb4c5e1c913449
--- /dev/null
+++ b/phase2/myosuite/soccer/trajectories/myosuite-soccer-i00-standard--sys8e3i.json
@@ -0,0 +1,223 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "myosuite-soccer-i00-standard--sys8e3i",
+ "taskKey": "myosuite/soccer",
+ "dag_sha256": "6017c07ff7b03b931ba35675b387457077288bf58ccba56e403c929a63483718",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "sim_error": 0,
+   "grader_error": 0,
+   "no_verdict": 0,
+   "n_actions": 219,
+   "sim_steps": 219,
+   "physics_steps": 2190,
+   "steps": 554,
+   "calls": 50,
+   "episodes": 3,
+   "resets": 2,
+   "reset_budget": 10,
+   "sim_steps_total": 637,
+   "elapsed_s": 2737.8,
+   "video_rendered": 1,
+   "task_failure": 1,
+   "timeout": 0,
+   "instance_mismatch": 0,
+   "gave_up": 0,
+   "sim_crashed_on_call": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_joint_dev": 0.0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.844,
+   "peak_processes": 15,
+   "peak_disk_gb": 0.383,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/myosuite-soccer-i00-standard--sys8e3i/replay.mp4",
+  "duration_s": 8.8,
+  "fps": 25.0,
+  "frames": 220,
+  "sha256": "4cdd5f5e46feb03ad77451469aa1f04159748b05c0c1165003407ed81ca35470"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "ball_in_goal": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0.0 s 足球位于画面右下方起球处，白色球门位于左上方远端，初态未进球。"
+     },
+     {
+      "t": 5.28,
+      "log_step": null,
+      "text_zh": "观察：视频约 4.4–5.3 s 模型前倾下落，身体与足球在画面中局部重叠；视频约 5.3–6.0 s 球抬起并向球门方向离开原位，模型未恢复站立。推断：模型碰动了足球，但触球身体段及首次接触的准确时刻不能确定。"
+     },
+     {
+      "t": 8.76,
+      "log_step": 142,
+      "text_zh": "观察：视频 7.8–8.76 s 足球仍在起球区域附近、远未到达球门线，模型已伏倒；全片未见球到达门线。log step 142 最后一次重置后的执行返回 done=True、失败原因为 fallen 或未进球超时、仿真终点 2.19 s；评分器 live_success=0、graded_by_live=1，与视频一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频约 2.5–4.4 s 模型的腿从画面右缘进入球附近；视频约 4.4–5.3 s 球与下落的身体局部重叠，视频约 5.3–6.0 s 球抬起并向球门方向离开。推断：模型某处身体碰动了足球；画面不能区分触球的是腿、脚还是躯干，也不能确定首次触球的准确时刻。候选接触名为“击打”，词表无撞击或踢击对应项，使用其他。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频约 3.4–4.6 s 可见模型一侧脚/腿落在球场地面，身体同时向球前倾；随后身体继续下沉并伏倒。推断：该脚/腿在此前倾阶段承担支撑；视频 0–2.0 s 模型大部在画面外，不能确认更早的支撑起点。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_cost",
+   "value": null,
+   "unit": "s",
+   "threshold": 10,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_in_goal",
+   "t": 4.8,
+   "log_step": 142,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 4.6–5.3 s，模型在接近并撞动足球时身体持续下沉、前扑，随后始终未站起；球到视频 8.76 s 仍远离球门。log step 142 为最后一次重置后的执行，服务记录仿真 2.19 s 失败结束。推断：踢球入门这一步中的失衡未被修复，故断点归于 ball_in_goal；视频 4.8 s 为倒伏过程的粗略导航锚点。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 4.8,
+    "log_step": 142,
+    "evidence_zh": "观察：视频 3.4–4.6 s 模型明显向前折下，视频约 4.8–5.3 s 躯干降至接近地面，之后直至视频 8.76 s 仍伏倒，未恢复足部支撑下的站立；此时足球尚远未到门线。log step 142 服务在仿真 2.19 s 返回失败，早于仿真十秒上限。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "log_step": 128,
+    "evidence_zh": "观察：log step 128 的执行脚本把离线搜索所得肌肉基线与高斯脉冲展开为动作表，一次性开环提交，再用 get_obs 维持末次命令观察；log step 140、142 记录选用该参数、重置后提交 205×290 动作表并以仿真 2.19 s 失败结束。视频约 4.8–5.3 s 显示该动作使模型前扑，仅撞动足球，未能兼顾支撑与击球。推断：实际肌肉控制序列未产生完成射门所需的协调动作，根因归控制；证据不足以细分为反馈控制、逆解或某一参数错误。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "无状态日志，无法精确核验逐步骨盆高度、撞球身体段及接触冲量；低分辨率画面仅能观察到下落身体与球局部重叠、随后球离开，据此推断模型碰动足球，无法确定首次触球的准确时刻。视频早段模型大部在画面外，不能确认视频 2.0 s 已有足部支撑；视频约 3.4 s 起可见脚/腿落地。日志里的离线搜索和前两个在线回合不作为本次节点进度；仿真 2.19 s 与视频 8.76 s 为不同时间轴。评分器未报告 time_cost，故该指标留空，不能把仿真终止时间当成进球用时。",
+  "unmet": {
+   "ball_in_goal": false
+  }
+ },
+ "summary_zh": "最后一次在线回合撞动了足球，但球始终未到达门线，唯一目标“踢球入门”未完成（0/1）。模型在视频约 4.8–5.3 s 前扑倒伏，直到视频 8.76 s 未恢复；执行日志支持肌肉控制序列未能协调支撑与击球，根因归控制。无状态日志，具体撞球身体段、骨盆阈值触发时刻及控制子类型不能精确确定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点 ball_in_goal、平衡现象与控制根因与视频及评分器结果一致；唯一问题是 contact_actual 把撞球部位断为“躯干前侧”、支撑起点写 2.0 s，画面不足以支持这些观察。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "节点未达成、断点 ball_in_goal、平衡现象、控制根因与视频及评分器结果一致；第1轮的接触断言已按证据修正，未发现新错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "codex",
+  "model": "gpt-6.1-sol",
+  "effort": "high",
+  "created": "2026-10-30T04:13:45Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--sys8e3i/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--sys8e3i/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--sys8e3i/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/myosuite-soccer-i00-standard--sys8e3i/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ba3a706ba1a80b9669fb08384f9fe01814042a2fb05688e55a15a113607e1b69",
+   "video_sha256": "4cdd5f5e46feb03ad77451469aa1f04159748b05c0c1165003407ed81ca35470",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  }
+ }
+}