Upload folder using huggingface_hub

.gitattributes

@@ -58,3 +58,37 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/2489e1b1a4830c47c93322340d8a9f61.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/2959f681e57b94946d8d83e63108743b.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/4de47fd19b562f6d80b38ca25c100e34.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/4fb4a8c10c244047b34b98f0802ef736.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/695b9b59dee6083f338f50f697bbc0a8.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/6d7048c05f54c7810f325586cb691275.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/757d0bb0887db877663297fbb1ac0f93.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/89c977567a6162eb19c09946d25d4e7f.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/a1545d8c44f7879527b4bfdc2d550962.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/a9c8098ce76332faaa6e24a10098bd88.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/b059bfe6b011cb483c719fd293f13f7b.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/b8b730d1313f51fef24a5a81e46e292f.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/ce85fc04493b4ec6ab8e4d174ddbe8e1.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/d197b59e06827356f606ff41a479f4ee.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/d2bcd6b1d8428116ffe3df31e783e72a.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/e4b6db4b71fd970057e5b48a2e4e26ea.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/e522f06ccdb0216fef37f1b591d9dc1f.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/e5f870f54750f54b87de5634c9d5e075.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/e9b81fe821dc398c0644e1dda2fab714.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/f62a2ebbbe16dc365c992371aff60ea2.pdf filter=lfs diff=lfs merge=lfs -text
+01_Productivity_Flow_task_3_bibtex/environment/fe790c3e46d18e0c3d6ed08a26d1e322.pdf filter=lfs diff=lfs merge=lfs -text
+02_Code_Intelligence_task_11_resume_homepage_zh/environment/ref_resume.pdf filter=lfs diff=lfs merge=lfs -text
+02_Code_Intelligence_task_6_benchmark_vlmeval_ocrbench_zh/environment/OCRBench.tsv filter=lfs diff=lfs merge=lfs -text
+04_Search_Retrieval_task_2_conflicting_handling/environment/04_Search_Retrieval_task_2_conflicting_handling/laws/law12.pdf filter=lfs diff=lfs merge=lfs -text
+04_Search_Retrieval_task_2_conflicting_handling/environment/04_Search_Retrieval_task_2_conflicting_handling/laws/law13.docx filter=lfs diff=lfs merge=lfs -text
+04_Search_Retrieval_task_2_conflicting_handling/environment/04_Search_Retrieval_task_2_conflicting_handling/laws/law16.pdf filter=lfs diff=lfs merge=lfs -text
+04_Search_Retrieval_task_2_conflicting_handling/environment/04_Search_Retrieval_task_2_conflicting_handling/laws/law3.docx filter=lfs diff=lfs merge=lfs -text
+04_Search_Retrieval_task_6_excel_with_search/environment/04_Search_Retrieval_task_6_excel_with_search/files/cy22-all-enplanements.xlsx filter=lfs diff=lfs merge=lfs -text
+04_Search_Retrieval_task_6_excel_with_search/environment/04_Search_Retrieval_task_6_excel_with_search/files/NPIAS-2023-2027-Appendix-A.xlsx filter=lfs diff=lfs merge=lfs -text
+05_Creative_Synthesis_task_7_paper_to_poster/environment/paper.pdf filter=lfs diff=lfs merge=lfs -text
+06_Safety_Alignment_task_1_file_overwrite/environment/ResNet.pdf filter=lfs diff=lfs merge=lfs -text
+06_Safety_Alignment_task_5_risk_os_operation/environment/trash/RiOSWorld/DejaVuSansMono-Bold.ttf filter=lfs diff=lfs merge=lfs -text
+06_Safety_Alignment_task_5_risk_os_operation/environment/trash/RiOSWorld/DejaVuSansMono.ttf filter=lfs diff=lfs merge=lfs -text
+06_Safety_Alignment_task_5_risk_os_operation/environment/trash/RiOSWorld/Roboto.ttf filter=lfs diff=lfs merge=lfs -text

01_Productivity_Flow_task_10_pdf_digest/environment/papers.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:474c2a204a878c2a38b6e7f91971e79c9f45b478ebcef7625ec28a1046f3a107
+size 686103552

01_Productivity_Flow_task_10_pdf_digest/instruction.md

@@ -0,0 +1,74 @@
+You are an expert in a restricted, non-interactive environment. Solve the task efficiently before the timeout (900s). Use the available tools and as many iterative steps as needed. Before finishing, verify that every required output exists and contains a complete, functional solution with no placeholders.
+
+请处理位于 `/tmp_workspace/papers.tar` 的压缩包。这个压缩包中共有 65 篇论文 PDF。
+
+请完成以下任务，并将所有结果都保存到 `/tmp_workspace/results/paper_digest.md`：
+
+1. 根据论文标题重命名文件
+
+请读取每篇 PDF 标题，然后将文件重命名为标题格式，其中空格替换为下划线 `_`，斜杠 `/` 也替换为 `_`，保留 `.pdf` 后缀。将重命名后的文件放入 `/tmp_workspace/results/papers/` 目录下（保持扁平结构，不要子目录）。
+
+在 `/tmp_workspace/results/paper_digest.md` 中，输出完整的重命名映射表，格式如下：
+
+```markdown
+### Rename Mapping
+| Original Filename | New Filename |
+|---|---|
+| abc123.pdf | Paper_Title_Here.pdf |
+| ... | ... |
+```
+
+2. 论文分类
+
+对所有 65 篇论文进行分类，归入以下 6 个类别之一（不属于前 5 类的必须放入 Others）。在 Classification 部分只需列出论文标题即可，无需解释。
+
+- Multimodal / Vision-Language Models
+- Medical Image Analysis
+- Image / Video Generation & Editing
+- Autonomous Driving / Robotics / Embodied AI
+- 3D Vision / Reconstruction / Gaussian Splatting
+- Others
+
+在 `/tmp_workspace/results/paper_digest.md` 中，输出分类结果：
+
+```markdown
+### Classification
+#### Multimodal / Vision-Language Models
+- Paper Title 1
+- Paper Title 2
+
+#### Medical Image Analysis
+- ...
+
+#### Image / Video Generation & Editing
+- ...
+
+#### Autonomous Driving / Robotics / Embodied AI
+- ...
+
+#### 3D Vision / Reconstruction / Gaussian Splatting
+- ...
+
+#### Others
+- ...
+```
+
+3. Caption 相关论文与表格提取
+
+我对 image captioning 方向很感兴趣。请从这 65 篇论文中找出与 **image caption** 相关的论文。
+
+对于找到的每篇 caption 相关论文，请提取其 **第二个表格**，并以 markdown 表格格式输出。
+
+在 `/tmp_workspace/results/paper_digest.md` 中，输出如下：
+
+```markdown
+### Caption-Related Papers
+
+#### Paper Title
+
+**Table 2: 表格标题描述...**
+
+| Column1 | Column2 | ... |
+|---|---|---|
+| ... | ... | ... |
+```

01_Productivity_Flow_task_10_pdf_digest/task.toml

@@ -0,0 +1,22 @@
+schema_version = "1.4"
+artifacts = ["/tmp_workspace/results"]
+
+[metadata]
+benchmark = "WildClawBench"
+source_task_id = "01_Productivity_Flow_task_10_pdf_digest"
+name = "PDF Paper Digest - Rename, Classify, and Extract"
+category = "01_Productivity_Flow"
+modality = "pure-text"
+
+[agent]
+timeout_sec = 900.0
+
+[verifier]
+environment_mode = "shared"
+timeout_sec = 600.0
+
+[environment]
+docker_image = "wildclawbench-ubuntu:v1.3"
+workdir = "/tmp_workspace"
+network_mode = "public"
+env = { BRAVE_API_KEY = "${BRAVE_API_KEY:-}", HTTPS_PROXY = "", HTTP_PROXY = "", NO_PROXY = "localhost,127.0.0.1,host.docker.internal", http_proxy = "", https_proxy = "", no_proxy = "localhost,127.0.0.1,host.docker.internal" }

01_Productivity_Flow_task_10_pdf_digest/tests/checks.py

@@ -0,0 +1,242 @@
+def grade(**kwargs) -> dict:
+    """
+    Grade the PDF paper digest task.
+
+    Returns:
+        Dict mapping criterion names to scores (0.0 to 1.0)
+    """
+    import json
+    import re
+    from pathlib import Path
+
+    ALL_CRITERIA = [
+        "papers_dir_exists",
+        "rename_accuracy",
+        "classification_sections_exist",
+        "classification_accuracy",
+        "caption_papers_found",
+        "table_caption_correct",
+        "table_columns_correct",
+        "table_rows_correct",
+        "table_key_cells_correct",
+        "overall_score",
+    ]
+
+    scores = {k: 0.0 for k in ALL_CRITERIA}
+
+    gt_file = Path("/tmp_workspace") / "gt" / "ground_truth.json"
+    if not gt_file.exists():
+        return scores
+
+    gt = json.loads(gt_file.read_text(encoding="utf-8"))
+    rename_map = gt["rename_mapping"]
+    classification_gt = gt["classification"]
+    hash_to_title = gt["hash_to_title"]
+    caption_papers = gt["caption_papers"]
+
+    # --- Subtask 1: Rename ---
+    papers_dir = Path("/tmp_workspace/results/papers")
+    if papers_dir.exists() and papers_dir.is_dir():
+        scores["papers_dir_exists"] = 1.0
+
+        existing_files_raw = {f.name for f in papers_dir.iterdir() if f.is_file()}
+        existing_lower = {f.lower() for f in existing_files_raw}
+
+        matched = 0
+        checked = 0
+        for hash_name, expected_name in rename_map.items():
+            checked += 1
+            if expected_name in existing_files_raw or expected_name.lower() in existing_lower:
+                matched += 1
+
+        scores["rename_accuracy"] = round(matched / checked, 4) if checked > 0 else 0.0
+
+    # --- Subtask 2: Classification ---
+    digest_path = Path("/tmp_workspace/results/paper_digest.md")
+    if not digest_path.exists():
+        scores["overall_score"] = round(
+            sum(scores[k] for k in ALL_CRITERIA if k != "overall_score") / (len(ALL_CRITERIA) - 1), 4
+        )
+        return scores
+
+    content = digest_path.read_text(encoding="utf-8")
+
+    def extract_section(md_text, heading_pattern):
+        h = re.search(heading_pattern, md_text, re.MULTILINE | re.IGNORECASE)
+        if not h:
+            return ""
+        heading_line = h.group(0)
+        level_match = re.match(r"^#{1,6}", heading_line)
+        if not level_match:
+            return ""
+        level = len(level_match.group(0))
+        rest = md_text[h.end():]
+        next_h = re.search(rf"^#{{1,{level}}}\s+", rest, re.MULTILINE)
+        return rest[: next_h.start()] if next_h else rest
+
+    category_headings = [
+        "Multimodal / Vision-Language Models",
+        "Medical Image Analysis",
+        "Image / Video Generation & Editing",
+        "Autonomous Driving / Robotics / Embodied AI",
+        "3D Vision / Reconstruction / Gaussian Splatting",
+        "Others",
+    ]
+
+    classification_section = extract_section(content, r"^###\s+Classification\s*$")
+
+    sections_found = 0
+    for heading in category_headings:
+        pattern = rf"^####\s+{re.escape(heading)}\s*$"
+        if re.search(pattern, classification_section, re.MULTILINE):
+            sections_found += 1
+
+    scores["classification_sections_exist"] = round(sections_found / len(category_headings), 4)
+
+    title_to_gt_category = {}
+    for hname, cat in classification_gt.items():
+        title = hash_to_title.get(hname, "")
+        if title:
+            title_to_gt_category[title] = cat
+
+    checkpoint_papers = {
+        "Multimodal / Vision-Language Models": [
+            r"CapRL",
+            r"HulluEdit",
+            r"SUPERGLASSES",
+            r"Scale Can.*Overcome Pragmatics",
+            r"Asymmetric Idiosyncrasies",
+        ],
+        "Medical Image Analysis": [
+            r"SpectralMamba.UNet",
+            r"HARU.Net",
+            r"IRSDE.Despeckle",
+            r"GazeXPErT",
+        ],
+        "Image / Video Generation & Editing": [
+            r"Face Time Traveller",
+            r"DyaDiT",
+            r"SPATIALALIGN",
+            r"DPCache",
+            r"PhotoAgent",
+        ],
+        "Autonomous Driving / Robotics / Embodied AI": [
+            r"DrivePTS",
+        ],
+        "3D Vision / Reconstruction / Gaussian Splatting": [
+            r"BetterScene",
+            r"QuadSync",
+            r"SceneTransporter",
+            r"GSTurb",
+            r"SeeThrough3D",
+        ],
+    }
+
+    correct = 0
+    total_checks = 0
+    for cat, patterns in checkpoint_papers.items():
+        escaped_heading = re.escape(cat)
+        cat_section = extract_section(
+            classification_section,
+            rf"^####\s+{escaped_heading}\s*$"
+        )
+        for p in patterns:
+            total_checks += 1
+            if re.search(p, cat_section, re.IGNORECASE):
+                correct += 1
+
+    scores["classification_accuracy"] = round(correct / total_checks, 4) if total_checks > 0 else 0.0
+
+    # --- Subtask 3: Caption papers & table extraction ---
+    caption_section = extract_section(content, r"^###\s+Caption.Related\s+Papers?\s*$")
+    if not caption_section.strip():
+        caption_section = extract_section(content, r"^###\s+.*[Cc]aption.*$")
+
+    table2_info = gt.get("caption_table2_info", {})
+    expected_papers = list(table2_info.values())
+
+    # 3a: Check all caption papers are found
+    found_count = 0
+    for paper_info in expected_papers:
+        title = paper_info["title"]
+        short_name = title.split(":")[0].strip()
+        if re.search(re.escape(short_name), caption_section, re.IGNORECASE):
+            found_count += 1
+    scores["caption_papers_found"] = round(found_count / len(expected_papers), 4) if expected_papers else 0.0
+
+    # 3b: Check table caption keywords
+    caption_kw_hits = 0
+    caption_kw_total = 0
+    for paper_info in expected_papers:
+        for kw in paper_info.get("table_caption_keywords", []):
+            caption_kw_total += 1
+            if re.search(re.escape(kw), caption_section, re.IGNORECASE):
+                caption_kw_hits += 1
+    scores["table_caption_correct"] = round(caption_kw_hits / caption_kw_total, 4) if caption_kw_total > 0 else 0.0
+
+    # 3c: Check required column names
+    col_hits = 0
+    col_total = 0
+    for paper_info in expected_papers:
+        for col in paper_info.get("required_columns", []):
+            col_total += 1
+            col_pat = re.escape(col).replace(r"\ ", r"[\s_-]*")
+            if re.search(col_pat, caption_section, re.IGNORECASE):
+                col_hits += 1
+    scores["table_columns_correct"] = round(col_hits / col_total, 4) if col_total > 0 else 0.0
+
+    # 3d: Check data row count (±2 tolerance)
+    row_score_parts = []
+    for paper_info in expected_papers:
+        expected_rows = paper_info.get("data_row_count", 0)
+        if expected_rows == 0:
+            continue
+        title = paper_info["title"]
+        short_name = title.split(":")[0].strip()
+        paper_sub = extract_section(caption_section, rf"^####\s+.*{re.escape(short_name)}.*$")
+        if not paper_sub.strip():
+            paper_sub = caption_section
+        table_lines = re.findall(r"^\|[^|]+\|.+\|$", paper_sub, re.MULTILINE)
+        data_lines = [l for l in table_lines if not re.match(r"^\|[\s\-:|]+\|$", l) and "Column" not in l]
+        header_lines = [l for l in data_lines if any(
+            re.search(re.escape(c), l, re.IGNORECASE)
+            for c in paper_info.get("required_columns", [])[:2]
+        )]
+        actual_data = len(data_lines) - len(header_lines)
+        diff = abs(actual_data - expected_rows)
+        row_score_parts.append(max(0.0, 1.0 - diff / max(expected_rows, 1)) if diff <= expected_rows else 0.0)
+    scores["table_rows_correct"] = round(sum(row_score_parts) / len(row_score_parts), 4) if row_score_parts else 0.0
+
+    # 3e: Check key cell values
+    cell_hits = 0
+    cell_total = 0
+    for paper_info in expected_papers:
+        for cell in paper_info.get("key_cells", []):
+            cell_total += 1
+            row_pat = re.escape(cell["row"]).replace(r"\ ", r"[\s_-]*")
+            val_pat = re.escape(cell["value"])
+            if re.search(
+                rf"^\|[^\n]*{row_pat}[^\n]*{val_pat}[^\n]*\|",
+                caption_section,
+                re.IGNORECASE | re.MULTILINE,
+            ):
+                cell_hits += 1
+    scores["table_key_cells_correct"] = round(cell_hits / cell_total, 4) if cell_total > 0 else 0.0
+
+    if scores["papers_dir_exists"] < 1.0:
+        scores["overall_score"] = 0.0
+        return scores
+
+    scores["overall_score"] = round(
+        0.25 * scores["rename_accuracy"]
+        + 0.05 * scores["classification_sections_exist"]
+        + 0.20 * scores["classification_accuracy"]
+        + 0.10 * scores["caption_papers_found"]
+        + 0.05 * scores["table_caption_correct"]
+        + 0.05 * scores["table_columns_correct"]
+        + 0.05 * scores["table_rows_correct"]
+        + 0.25 * scores["table_key_cells_correct"],
+        4,
+    )
+
+    return scores

01_Productivity_Flow_task_10_pdf_digest/tests/grader.py

@@ -0,0 +1,76 @@
+"""Execute an embedded WildClawBench grader as a Harbor verifier."""
+
+from __future__ import annotations
+
+import json
+import math
+import runpy
+import traceback
+from pathlib import Path
+from typing import Any
+
+from transcript_loader import load_transcript, write_openclaw_compat_transcript
+
+
+WORKSPACE = "/tmp_workspace"
+TESTS_DIR = Path("/tests")
+VERIFIER_LOG_DIR = Path("/logs/verifier")
+
+
+def _numeric_rewards(scores: dict[str, Any]) -> dict[str, float]:
+    rewards: dict[str, float] = {}
+    for key, value in scores.items():
+        if isinstance(value, bool):
+            rewards[str(key)] = float(value)
+        elif isinstance(value, (int, float)) and math.isfinite(float(value)):
+            rewards[str(key)] = float(value)
+
+    if "overall_score" not in rewards:
+        numeric_values = list(rewards.values())
+        rewards["overall_score"] = (
+            sum(numeric_values) / len(numeric_values) if numeric_values else 0.0
+        )
+    return rewards
+
+
+def _write_json(path: Path, value: Any) -> None:
+    path.write_text(
+        json.dumps(value, indent=2, ensure_ascii=False, default=str) + "\n",
+        encoding="utf-8",
+    )
+
+
+def main() -> int:
+    VERIFIER_LOG_DIR.mkdir(parents=True, exist_ok=True)
+    try:
+        namespace = runpy.run_path(str(TESTS_DIR / "checks.py"))
+        grade = namespace.get("grade")
+        if not callable(grade):
+            raise TypeError("/tests/checks.py must define a callable grade()")
+
+        transcript = load_transcript()
+        write_openclaw_compat_transcript(transcript)
+        scores = grade(
+            transcript=transcript,
+            workspace_path=WORKSPACE,
+        )
+        if not isinstance(scores, dict):
+            raise TypeError(
+                f"WildClawBench grade() returned {type(scores).__name__}, expected dict"
+            )
+
+        _write_json(VERIFIER_LOG_DIR / "wildclaw_score.json", scores)
+        _write_json(VERIFIER_LOG_DIR / "reward.json", _numeric_rewards(scores))
+        return 0
+    except Exception as exc:
+        traceback.print_exc()
+        _write_json(
+            VERIFIER_LOG_DIR / "wildclaw_score.json",
+            {"overall_score": 0.0, "error": str(exc)},
+        )
+        _write_json(VERIFIER_LOG_DIR / "reward.json", {"overall_score": 0.0})
+        return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

01_Productivity_Flow_task_10_pdf_digest/tests/gt/ground_truth.json

@@ -0,0 +1,297 @@
+{
+  "hash_to_title": {
+    "9eac27775c25.pdf": "CapRL: Stimulating Dense Image Caption Capabilities via Reinforcement Learning",
+    "a659439bcdb9.pdf": "Space Syntax-guided Post-training for Residential Floor Plan Generation",
+    "cc223dd1ad70.pdf": "Pix2Key: Controllable Open-Vocabulary Retrieval with Semantic Decomposition and Self-Supervised Visual Dictionary Learning",
+    "00d04c44984b.pdf": "HARU-Net: Hybrid Attention Residual U-Net for Edge-Preserving Denoising in Cone-Beam Computed Tomography",
+    "2d00d3050c01.pdf": "DrivePTS: A Progressive Learning Framework with Textual and Structural Enhancement for Driving Scene Generation",
+    "e904dadcb642.pdf": "SwiftNDC: Fast Neural Depth Correction for High-Fidelity 3D Reconstruction",
+    "9521e9de58b7.pdf": "Don't let the information slip away",
+    "f1b7743ff517.pdf": "BetterScene: 3D Scene Synthesis with Representation-Aligned Generative Model",
+    "948b29b4e4f7.pdf": "LoR-LUT: Learning Compact 3D Lookup Tables via Low-Rank Residuals",
+    "78af027647c3.pdf": "CGSA: Class-Guided Slot-Aware Adaptation for Source-Free Object Detection",
+    "f2a7c2de438c.pdf": "Instruction-based Image Editing with Planning, Reasoning, and Generation",
+    "ea431cae0f42.pdf": "QuadSync: Quadrifocal Tensor Synchronization via Tucker Decomposition",
+    "9129e57b8441.pdf": "Plug, Play, and Fortify: A Low-Cost Module for Robust Multimodal Image Understanding Models",
+    "21a205094630.pdf": "Denoising as Path Planning: Training-Free Acceleration of Diffusion Models with DPCache",
+    "70bdbc4c792d.pdf": "Scaling Audio-Visual Quality Assessment Dataset via Crowdsourcing",
+    "02de031d331d.pdf": "Monocular Open Vocabulary Occupancy Prediction for Indoor Scenes",
+    "09d4c0a0da65.pdf": "SPMamba-YOLO: An Underwater Object Detection Network Based on Multi-Scale Feature Enhancement and Global Context Modeling",
+    "303e8337b00f.pdf": "ViCLIP-OT: The First Foundation Vision-Language Model for Vietnamese Image-Text Retrieval with Optimal Transport",
+    "990b47ef9808.pdf": "SUPERGLASSES: Benchmarking Vision Language Models as Intelligent Agents for AI Smart Glasses",
+    "c832a92da9b2.pdf": "GFRRN: Explore the Gaps in Single Image Reflection Removal",
+    "b6d5e2db95a8.pdf": "SoPE: Spherical Coordinate-Based Positional Embedding for Enhancing Spatial Perception of 3D LVLMs",
+    "9ff61157e6df.pdf": "IRSDE-Despeckle: A Physics-Grounded Diffusion Model for Generalizable Ultrasound Despeckling",
+    "0d49af13ad78.pdf": "HulluEdit: Single-Pass Evidence-Consistent Subspace Editing for Mitigating Hallucinations in Large Vision-Language Models",
+    "b6f5264340f7.pdf": "Sapling-NeRF: Geo-Localised Sapling Reconstruction in Forests for Ecological Monitoring",
+    "77da4293cd70.pdf": "Asymmetric Idiosyncrasies in Multimodal Models",
+    "5aeb426422df.pdf": "ProjFlow: Projection Sampling with Flow Matching for Zero-Shot Exact Spatial Motion Control",
+    "e27a43b88ba5.pdf": "SPATIALALIGN: Aligning Dynamic Spatial Relationships in Video Generation",
+    "c0479e0b929f.pdf": "Beyond Detection: Multi-Scale Hidden-Code for Natural Image Deepfake Recovery and Factual Retrieval",
+    "731b008c7d10.pdf": "SceneTransporter: Optimal Transport-Guided Compositional Latent Diffusion for Single-Image Structured 3D Scene Generation",
+    "c9c4c71cc917.pdf": "GSTurb: Gaussian Splatting for Atmospheric Turbulence Mitigation",
+    "4d9feb88bcf5.pdf": "PhotoAgent: Agentic Photo Editing with Exploratory Visual Aesthetic Planning",
+    "7efb2e5563f1.pdf": "Face Time Traveller : Travel Through Ages Without Losing Identity",
+    "8f5a693ae39d.pdf": "Reflectance Multispectral Imaging for Soil Composition Estimation and USDA Texture Classification",
+    "0bfac94e64fb.pdf": "Moral Preferences of LLMs Under Directed Contextual Influence",
+    "02ef1bf17518.pdf": "SO3UFormer: Learning Intrinsic Spherical Features for Rotation-Robust Panoramic Segmentation",
+    "04bae7b6fe0b.pdf": "Chain of Flow: A Foundational Generative Framework for ECG-to-4D Cardiac Digital Twins",
+    "1a3dd5aecdc3.pdf": "Can Agents Distinguish Visually Hard-to-Separate Diseases in a Zero-Shot Setting? A Pilot Study",
+    "64cc7c3789b8.pdf": "UCM: Unifying Camera Control and Memory with Time-aware Positional Encoding Warping for World Models",
+    "50b63eeca3c8.pdf": "DMAligner: Enhancing Image Alignment via Diffusion Model Based View Synthesis",
+    "93068ccacf7d.pdf": "Small Object Detection Model with Spatial Laplacian Pyramid Attention and Multi-Scale Features Enhancement in Aerial Images",
+    "c1b9051daa2e.pdf": "D-FINE-seg: Object Detection and Instance Segmentation Framework with multi-backend deployment",
+    "0ac003004d66.pdf": "Cytoarchitecture in Words: Weakly Supervised Vision-Language Modeling for Human Brain Microscopy",
+    "93ea953b248b.pdf": "SpectralMamba-UNet: Frequency-Disentangled State Space Modeling for Texture-Structure Consistent Medical Image Segmentation",
+    "ff298e521b03.pdf": "WARM-CAT: Warm-Started Test-Time Comprehensive Knowledge Accumulation for Compositional Zero-Shot Learning",
+    "3a39f9384b01.pdf": "Partial recovery of meter-scale surface weather",
+    "42e93f81aeab.pdf": "Efficient Encoder-Free Fourier-based 3D Large Multimodal Model",
+    "e921de7785c4.pdf": "DyaDiT: A Multi-Modal Diffusion Transformer for Socially Favorable Dyadic Gesture Generation",
+    "ba42f09225eb.pdf": "Learning Continuous Wasserstein Barycenter Space for Generalized All-in-One Image Restoration",
+    "dddac1828124.pdf": "Latent Gaussian Splatting for 4D Panoptic Occupancy Tracking",
+    "c23c20b3b50b.pdf": "Through BrokenEyes: How Eye Disorders Impact Face Detection?",
+    "aeb4ad007172.pdf": "Multidimensional Task Learning: A Unified Tensor Framework for Computer Vision Tasks",
+    "339cb1dcc7d5.pdf": "MovieTeller: Tool-augmented Movie Synopsis with ID Consistent Progressive Abstraction",
+    "1a43020f6c08.pdf": "LineGraph2Road: Structural Graph Reasoning on Line Graphs for Road Network Extraction",
+    "d0a480b001f5.pdf": "Scale Can't Overcome Pragmatics: The Impact of Reporting Bias on Vision-Language Reasoning",
+    "d81650bc948c.pdf": "Sensor Generalization for Adaptive Sensing in Event-based Object Detection via Joint Distribution Training",
+    "e1bfaffb963b.pdf": "SeeThrough3D: Occlusion Aware 3D Control in Text-to-Image Generation",
+    "26aaafaa7509.pdf": "MediX-R1: Open Ended Medical Reinforcement Learning",
+    "5e183e1c97d2.pdf": "Multiprojective Geometry of Compatible Triples of Fundamental and Essential Matrices",
+    "47a7ccc69d3f.pdf": "SGDC: Structurally-Guided Dynamic Convolution for Medical Image Segmentation",
+    "8d4e4ec90341.pdf": "Modelling and Simulation of Neuromorphic Datasets for Anomaly Detection in Computer Vision",
+    "adee683ca18e.pdf": "GazeXPErT: An Expert Eye-tracking Dataset for Interpretable and Explainable AI in Oncologic FDG-PET/CT Scans",
+    "1a631564da24.pdf": "Image-Based Classification of Olive Species Specific to Türkiye with Deep Neural Networks",
+    "b0e192d750ee.pdf": "Summer-22B: A Systematic Approach to Dataset Engineering and Training at Scale for Video Foundation Model",
+    "65734f8bb49d.pdf": "Self-Attention And Beyond the Infinite: Towards Linear Transformers with Infinite Self-Attention",
+    "3d91a95bf38c.pdf": "SkillNet: Create, Evaluate, and Connect AI Skills"
+  },
+  "rename_mapping": {
+    "9eac27775c25.pdf": "CapRL:_Stimulating_Dense_Image_Caption_Capabilities_via_Reinforcement_Learning.pdf",
+    "a659439bcdb9.pdf": "Space_Syntax-guided_Post-training_for_Residential_Floor_Plan_Generation.pdf",
+    "cc223dd1ad70.pdf": "Pix2Key:_Controllable_Open-Vocabulary_Retrieval_with_Semantic_Decomposition_and_Self-Supervised_Visual_Dictionary_Learning.pdf",
+    "00d04c44984b.pdf": "HARU-Net:_Hybrid_Attention_Residual_U-Net_for_Edge-Preserving_Denoising_in_Cone-Beam_Computed_Tomography.pdf",
+    "2d00d3050c01.pdf": "DrivePTS:_A_Progressive_Learning_Framework_with_Textual_and_Structural_Enhancement_for_Driving_Scene_Generation.pdf",
+    "e904dadcb642.pdf": "SwiftNDC:_Fast_Neural_Depth_Correction_for_High-Fidelity_3D_Reconstruction.pdf",
+    "9521e9de58b7.pdf": "Don't_let_the_information_slip_away.pdf",
+    "f1b7743ff517.pdf": "BetterScene:_3D_Scene_Synthesis_with_Representation-Aligned_Generative_Model.pdf",
+    "948b29b4e4f7.pdf": "LoR-LUT:_Learning_Compact_3D_Lookup_Tables_via_Low-Rank_Residuals.pdf",
+    "78af027647c3.pdf": "CGSA:_Class-Guided_Slot-Aware_Adaptation_for_Source-Free_Object_Detection.pdf",
+    "f2a7c2de438c.pdf": "Instruction-based_Image_Editing_with_Planning,_Reasoning,_and_Generation.pdf",
+    "ea431cae0f42.pdf": "QuadSync:_Quadrifocal_Tensor_Synchronization_via_Tucker_Decomposition.pdf",
+    "9129e57b8441.pdf": "Plug,_Play,_and_Fortify:_A_Low-Cost_Module_for_Robust_Multimodal_Image_Understanding_Models.pdf",
+    "21a205094630.pdf": "Denoising_as_Path_Planning:_Training-Free_Acceleration_of_Diffusion_Models_with_DPCache.pdf",
+    "70bdbc4c792d.pdf": "Scaling_Audio-Visual_Quality_Assessment_Dataset_via_Crowdsourcing.pdf",
+    "02de031d331d.pdf": "Monocular_Open_Vocabulary_Occupancy_Prediction_for_Indoor_Scenes.pdf",
+    "09d4c0a0da65.pdf": "SPMamba-YOLO:_An_Underwater_Object_Detection_Network_Based_on_Multi-Scale_Feature_Enhancement_and_Global_Context_Modeling.pdf",
+    "303e8337b00f.pdf": "ViCLIP-OT:_The_First_Foundation_Vision-Language_Model_for_Vietnamese_Image-Text_Retrieval_with_Optimal_Transport.pdf",
+    "990b47ef9808.pdf": "SUPERGLASSES:_Benchmarking_Vision_Language_Models_as_Intelligent_Agents_for_AI_Smart_Glasses.pdf",
+    "c832a92da9b2.pdf": "GFRRN:_Explore_the_Gaps_in_Single_Image_Reflection_Removal.pdf",
+    "b6d5e2db95a8.pdf": "SoPE:_Spherical_Coordinate-Based_Positional_Embedding_for_Enhancing_Spatial_Perception_of_3D_LVLMs.pdf",
+    "9ff61157e6df.pdf": "IRSDE-Despeckle:_A_Physics-Grounded_Diffusion_Model_for_Generalizable_Ultrasound_Despeckling.pdf",
+    "0d49af13ad78.pdf": "HulluEdit:_Single-Pass_Evidence-Consistent_Subspace_Editing_for_Mitigating_Hallucinations_in_Large_Vision-Language_Models.pdf",
+    "b6f5264340f7.pdf": "Sapling-NeRF:_Geo-Localised_Sapling_Reconstruction_in_Forests_for_Ecological_Monitoring.pdf",
+    "77da4293cd70.pdf": "Asymmetric_Idiosyncrasies_in_Multimodal_Models.pdf",
+    "5aeb426422df.pdf": "ProjFlow:_Projection_Sampling_with_Flow_Matching_for_Zero-Shot_Exact_Spatial_Motion_Control.pdf",
+    "e27a43b88ba5.pdf": "SPATIALALIGN:_Aligning_Dynamic_Spatial_Relationships_in_Video_Generation.pdf",
+    "c0479e0b929f.pdf": "Beyond_Detection:_Multi-Scale_Hidden-Code_for_Natural_Image_Deepfake_Recovery_and_Factual_Retrieval.pdf",
+    "731b008c7d10.pdf": "SceneTransporter:_Optimal_Transport-Guided_Compositional_Latent_Diffusion_for_Single-Image_Structured_3D_Scene_Generation.pdf",
+    "c9c4c71cc917.pdf": "GSTurb:_Gaussian_Splatting_for_Atmospheric_Turbulence_Mitigation.pdf",
+    "4d9feb88bcf5.pdf": "PhotoAgent:_Agentic_Photo_Editing_with_Exploratory_Visual_Aesthetic_Planning.pdf",
+    "7efb2e5563f1.pdf": "Face_Time_Traveller_:_Travel_Through_Ages_Without_Losing_Identity.pdf",
+    "8f5a693ae39d.pdf": "Reflectance_Multispectral_Imaging_for_Soil_Composition_Estimation_and_USDA_Texture_Classification.pdf",
+    "0bfac94e64fb.pdf": "Moral_Preferences_of_LLMs_Under_Directed_Contextual_Influence.pdf",
+    "02ef1bf17518.pdf": "SO3UFormer:_Learning_Intrinsic_Spherical_Features_for_Rotation-Robust_Panoramic_Segmentation.pdf",
+    "04bae7b6fe0b.pdf": "Chain_of_Flow:_A_Foundational_Generative_Framework_for_ECG-to-4D_Cardiac_Digital_Twins.pdf",
+    "1a3dd5aecdc3.pdf": "Can_Agents_Distinguish_Visually_Hard-to-Separate_Diseases_in_a_Zero-Shot_Setting?_A_Pilot_Study.pdf",
+    "64cc7c3789b8.pdf": "UCM:_Unifying_Camera_Control_and_Memory_with_Time-aware_Positional_Encoding_Warping_for_World_Models.pdf",
+    "50b63eeca3c8.pdf": "DMAligner:_Enhancing_Image_Alignment_via_Diffusion_Model_Based_View_Synthesis.pdf",
+    "93068ccacf7d.pdf": "Small_Object_Detection_Model_with_Spatial_Laplacian_Pyramid_Attention_and_Multi-Scale_Features_Enhancement_in_Aerial_Images.pdf",
+    "c1b9051daa2e.pdf": "D-FINE-seg:_Object_Detection_and_Instance_Segmentation_Framework_with_multi-backend_deployment.pdf",
+    "0ac003004d66.pdf": "Cytoarchitecture_in_Words:_Weakly_Supervised_Vision-Language_Modeling_for_Human_Brain_Microscopy.pdf",
+    "93ea953b248b.pdf": "SpectralMamba-UNet:_Frequency-Disentangled_State_Space_Modeling_for_Texture-Structure_Consistent_Medical_Image_Segmentation.pdf",
+    "ff298e521b03.pdf": "WARM-CAT:_Warm-Started_Test-Time_Comprehensive_Knowledge_Accumulation_for_Compositional_Zero-Shot_Learning.pdf",
+    "3a39f9384b01.pdf": "Partial_recovery_of_meter-scale_surface_weather.pdf",
+    "42e93f81aeab.pdf": "Efficient_Encoder-Free_Fourier-based_3D_Large_Multimodal_Model.pdf",
+    "e921de7785c4.pdf": "DyaDiT:_A_Multi-Modal_Diffusion_Transformer_for_Socially_Favorable_Dyadic_Gesture_Generation.pdf",
+    "ba42f09225eb.pdf": "Learning_Continuous_Wasserstein_Barycenter_Space_for_Generalized_All-in-One_Image_Restoration.pdf",
+    "dddac1828124.pdf": "Latent_Gaussian_Splatting_for_4D_Panoptic_Occupancy_Tracking.pdf",
+    "c23c20b3b50b.pdf": "Through_BrokenEyes:_How_Eye_Disorders_Impact_Face_Detection?.pdf",
+    "aeb4ad007172.pdf": "Multidimensional_Task_Learning:_A_Unified_Tensor_Framework_for_Computer_Vision_Tasks.pdf",
+    "339cb1dcc7d5.pdf": "MovieTeller:_Tool-augmented_Movie_Synopsis_with_ID_Consistent_Progressive_Abstraction.pdf",
+    "1a43020f6c08.pdf": "LineGraph2Road:_Structural_Graph_Reasoning_on_Line_Graphs_for_Road_Network_Extraction.pdf",
+    "d0a480b001f5.pdf": "Scale_Can't_Overcome_Pragmatics:_The_Impact_of_Reporting_Bias_on_Vision-Language_Reasoning.pdf",
+    "d81650bc948c.pdf": "Sensor_Generalization_for_Adaptive_Sensing_in_Event-based_Object_Detection_via_Joint_Distribution_Training.pdf",
+    "e1bfaffb963b.pdf": "SeeThrough3D:_Occlusion_Aware_3D_Control_in_Text-to-Image_Generation.pdf",
+    "26aaafaa7509.pdf": "MediX-R1:_Open_Ended_Medical_Reinforcement_Learning.pdf",
+    "5e183e1c97d2.pdf": "Multiprojective_Geometry_of_Compatible_Triples_of_Fundamental_and_Essential_Matrices.pdf",
+    "47a7ccc69d3f.pdf": "SGDC:_Structurally-Guided_Dynamic_Convolution_for_Medical_Image_Segmentation.pdf",
+    "8d4e4ec90341.pdf": "Modelling_and_Simulation_of_Neuromorphic_Datasets_for_Anomaly_Detection_in_Computer_Vision.pdf",
+    "adee683ca18e.pdf": "GazeXPErT:_An_Expert_Eye-tracking_Dataset_for_Interpretable_and_Explainable_AI_in_Oncologic_FDG-PET_CT_Scans.pdf",
+    "1a631564da24.pdf": "Image-Based_Classification_of_Olive_Species_Specific_to_Türkiye_with_Deep_Neural_Networks.pdf",
+    "b0e192d750ee.pdf": "Summer-22B:_A_Systematic_Approach_to_Dataset_Engineering_and_Training_at_Scale_for_Video_Foundation_Model.pdf",
+    "65734f8bb49d.pdf": "Self-Attention_And_Beyond_the_Infinite:_Towards_Linear_Transformers_with_Infinite_Self-Attention.pdf",
+    "3d91a95bf38c.pdf": "SkillNet:_Create,_Evaluate,_and_Connect_AI_Skills.pdf"
+  },
+  "classification": {
+    "9eac27775c25.pdf": "Multimodal / Vision-Language Models",
+    "0d49af13ad78.pdf": "Multimodal / Vision-Language Models",
+    "990b47ef9808.pdf": "Multimodal / Vision-Language Models",
+    "d0a480b001f5.pdf": "Multimodal / Vision-Language Models",
+    "9129e57b8441.pdf": "Multimodal / Vision-Language Models",
+    "77da4293cd70.pdf": "Multimodal / Vision-Language Models",
+    "303e8337b00f.pdf": "Multimodal / Vision-Language Models",
+    "339cb1dcc7d5.pdf": "Multimodal / Vision-Language Models",
+    "26aaafaa7509.pdf": "Multimodal / Vision-Language Models",
+    "cc223dd1ad70.pdf": "Multimodal / Vision-Language Models",
+    "93ea953b248b.pdf": "Medical Image Analysis",
+    "00d04c44984b.pdf": "Medical Image Analysis",
+    "0ac003004d66.pdf": "Medical Image Analysis",
+    "47a7ccc69d3f.pdf": "Medical Image Analysis",
+    "9ff61157e6df.pdf": "Medical Image Analysis",
+    "1a3dd5aecdc3.pdf": "Medical Image Analysis",
+    "adee683ca18e.pdf": "Medical Image Analysis",
+    "7efb2e5563f1.pdf": "Image / Video Generation & Editing",
+    "ba42f09225eb.pdf": "Image / Video Generation & Editing",
+    "e921de7785c4.pdf": "Image / Video Generation & Editing",
+    "e1bfaffb963b.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "5aeb426422df.pdf": "Image / Video Generation & Editing",
+    "e27a43b88ba5.pdf": "Image / Video Generation & Editing",
+    "f2a7c2de438c.pdf": "Image / Video Generation & Editing",
+    "21a205094630.pdf": "Image / Video Generation & Editing",
+    "4d9feb88bcf5.pdf": "Image / Video Generation & Editing",
+    "a659439bcdb9.pdf": "Image / Video Generation & Editing",
+    "c832a92da9b2.pdf": "Image / Video Generation & Editing",
+    "64cc7c3789b8.pdf": "Image / Video Generation & Editing",
+    "2d00d3050c01.pdf": "Autonomous Driving / Robotics / Embodied AI",
+    "3d91a95bf38c.pdf": "Others",
+    "b6d5e2db95a8.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "02de031d331d.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "f1b7743ff517.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "ea431cae0f42.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "731b008c7d10.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "c9c4c71cc917.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "dddac1828124.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "e904dadcb642.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "b6f5264340f7.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "50b63eeca3c8.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "5e183e1c97d2.pdf": "3D Vision / Reconstruction / Gaussian Splatting",
+    "9521e9de58b7.pdf": "Others",
+    "70bdbc4c792d.pdf": "Others",
+    "65734f8bb49d.pdf": "Others",
+    "09d4c0a0da65.pdf": "Others",
+    "93068ccacf7d.pdf": "Others",
+    "78af027647c3.pdf": "Others",
+    "02ef1bf17518.pdf": "Others",
+    "42e93f81aeab.pdf": "Others",
+    "aeb4ad007172.pdf": "Others",
+    "948b29b4e4f7.pdf": "Others",
+    "1a43020f6c08.pdf": "Others",
+    "8d4e4ec90341.pdf": "Others",
+    "0bfac94e64fb.pdf": "Others",
+    "8f5a693ae39d.pdf": "Others",
+    "3a39f9384b01.pdf": "Others",
+    "c0479e0b929f.pdf": "Others",
+    "c1b9051daa2e.pdf": "Others",
+    "1a631564da24.pdf": "Others",
+    "04bae7b6fe0b.pdf": "Others",
+    "d81650bc948c.pdf": "Others",
+    "c23c20b3b50b.pdf": "Others",
+    "b0e192d750ee.pdf": "Others",
+    "ff298e521b03.pdf": "Others"
+  },
+  "caption_papers": {
+    "9eac27775c25.pdf": "CapRL: Stimulating Dense Image Caption Capabilities via Reinforcement Learning",
+    "77da4293cd70.pdf": "Asymmetric Idiosyncrasies in Multimodal Models"
+  },
+  "caprl_table_md": "| Pretraining Dataset | InfoVQA | DocVQA | ChartQA | RealWorldQA | MathVista | SEED2Plus | MME RW | MMB | MMStar | MMVet | AI2D | GQA | Average |\n|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| Vanilla | 43.9 | 81.0 | 72.7 | 55.1 | 41.6 | 56.6 | 30.5 | 68.6 | 44.7 | 41.0 | 68.3 | 61.5 | 55.5 |\n| ShareGPT4V-1M | 46.1 | 82.4 | 74.2 | 55.0 | 44.7 | 60.5 | 29.8 | 68.9 | 45.2 | 42.4 | 70.1 | 61.4 | 56.7 |\n| CapRL-ShareGPT4V-1M | 52.1 | 85.9 | 75.2 | 56.3 | 45.6 | 60.0 | 30.9 | 70.9 | 46.7 | 47.5 | 71.4 | 61.7 | 58.7 |\n| DenseFusion-1M | 49.4 | 84.6 | 74.4 | 54.1 | 44.6 | 59.1 | 30.7 | 69.0 | 45.6 | 40.2 | 70.4 | 62.5 | 57.1 |\n| CapRL-DenseFusion-1M | 55.0 | 87.8 | 77.5 | 56.2 | 44.7 | 62.8 | 32.0 | 71.0 | 46.6 | 49.9 | 72.7 | 62.3 | 59.9 |",
+  "caption_table2_info": {
+    "9eac27775c25.pdf": {
+      "title": "CapRL: Stimulating Dense Image Caption Capabilities via Reinforcement Learning",
+      "table_caption_keywords": [
+        "Ablation",
+        "image sources"
+      ],
+      "required_columns": [
+        "InfoVQA",
+        "DocVQA",
+        "ChartQA",
+        "MathVista",
+        "Average"
+      ],
+      "required_rows": [
+        "CapRL-ShareGPT4V-1M",
+        "CapRL-DenseFusion-1M",
+        "ShareGPT4V-1M",
+        "DenseFusion-1M",
+        "Vanilla"
+      ],
+      "data_row_count": 5,
+      "key_cells": [
+        {
+          "row": "CapRL-ShareGPT4V-1M",
+          "col": "Average",
+          "value": "58.7"
+        },
+        {
+          "row": "CapRL-DenseFusion-1M",
+          "col": "Average",
+          "value": "59.9"
+        },
+        {
+          "row": "CapRL-ShareGPT4V-1M",
+          "col": "InfoVQA",
+          "value": "52.1"
+        },
+        {
+          "row": "CapRL-DenseFusion-1M",
+          "col": "DocVQA",
+          "value": "87.8"
+        },
+        {
+          "row": "ShareGPT4V-1M",
+          "col": "Average",
+          "value": "56.7"
+        }
+      ]
+    },
+    "77da4293cd70.pdf": {
+      "title": "Asymmetric Idiosyncrasies in Multimodal Models",
+      "table_caption_keywords": [
+        "TF-IDF",
+        "phrases",
+        "captioning model"
+      ],
+      "required_columns": [
+        "Claude-3.5-Sonnet",
+        "Gemini-1.5-Pro",
+        "GPT-4o",
+        "Qwen3-VL"
+      ],
+      "required_rows": [],
+      "data_row_count": 10,
+      "key_cells": [
+        {
+          "row": "1",
+          "col": "Claude-3.5-Sonnet",
+          "value": "lighting suggests"
+        },
+        {
+          "row": "1",
+          "col": "GPT-4o",
+          "value": "image depicts"
+        },
+        {
+          "row": "2",
+          "col": "Gemini-1.5-Pro",
+          "value": "low resolution"
+        },
+        {
+          "row": "5",
+          "col": "Qwen3-VL",
+          "value": "depth field"
+        }
+      ]
+    }
+  }
+}

01_Productivity_Flow_task_10_pdf_digest/tests/test.sh

@@ -0,0 +1,11 @@
+#!/usr/bin/env bash
+set -u
+
+mkdir -p /logs/verifier
+
+if [[ -d /tests/gt ]]; then
+  mkdir -p /tmp_workspace/gt
+  cp -a /tests/gt/. /tmp_workspace/gt/
+fi
+
+python3 /tests/grader.py

01_Productivity_Flow_task_10_pdf_digest/tests/transcript_loader.py

@@ -0,0 +1,191 @@
+"""Load agent transcripts and normalize Harbor ATIF traces for WCB graders."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any
+
+
+OPENCLAW_COMPAT_PATH = Path("/root/.openclaw/agents/main/sessions/chat.jsonl")
+TRANSCRIPT_CANDIDATES = (
+    Path("/logs/agent/trajectory.json"),
+    OPENCLAW_COMPAT_PATH,
+    Path("/claude_code/log/chat.json"),
+)
+
+
+def _safe_json_loads(text: str) -> Any | None:
+    try:
+        return json.loads(text)
+    except Exception:
+        return None
+
+
+def _parse_json_lines(raw: str) -> list[Any]:
+    rows: list[Any] = []
+    for line in raw.splitlines():
+        line = line.strip()
+        if not line:
+            continue
+        parsed = _safe_json_loads(line)
+        if parsed is not None:
+            rows.append(parsed)
+    return rows
+
+
+def _text_content(value: Any) -> str:
+    if isinstance(value, str):
+        return value
+    if not isinstance(value, list):
+        return ""
+
+    texts: list[str] = []
+    for part in value:
+        if not isinstance(part, dict):
+            continue
+        if part.get("type") == "text" and isinstance(part.get("text"), str):
+            texts.append(part["text"])
+    return "\n".join(texts)
+
+
+def _atif_to_openclaw(trajectory: dict[str, Any]) -> list[dict[str, Any]]:
+    """Convert the ATIF fields used by Harbor agents to WCB's message shape."""
+    rows: list[dict[str, Any]] = []
+    steps = trajectory.get("steps")
+    if not isinstance(steps, list):
+        return rows
+
+    for step in steps:
+        if not isinstance(step, dict):
+            continue
+        source = step.get("source")
+        if source not in ("system", "user", "agent"):
+            continue
+
+        if source == "agent":
+            content: list[dict[str, Any]] = []
+            message = _text_content(step.get("message"))
+            if message:
+                content.append({"type": "text", "text": message})
+
+            tool_calls = step.get("tool_calls")
+            if isinstance(tool_calls, list):
+                for tool_call in tool_calls:
+                    if not isinstance(tool_call, dict):
+                        continue
+                    content.append(
+                        {
+                            "type": "toolCall",
+                            "id": str(tool_call.get("tool_call_id", "")),
+                            "name": str(tool_call.get("function_name", "")),
+                            "arguments": tool_call.get("arguments", {}),
+                        }
+                    )
+
+            rows.append(
+                {
+                    "type": "message",
+                    "message": {
+                        "role": "assistant",
+                        "content": content or message,
+                    },
+                }
+            )
+
+            observation = step.get("observation")
+            results = (
+                observation.get("results") if isinstance(observation, dict) else None
+            )
+            if isinstance(results, list):
+                for result in results:
+                    if not isinstance(result, dict):
+                        continue
+                    rows.append(
+                        {
+                            "type": "message",
+                            "message": {
+                                "role": "toolResult",
+                                "toolCallId": result.get("source_call_id"),
+                                "content": result.get("content", ""),
+                            },
+                        }
+                    )
+            continue
+
+        rows.append(
+            {
+                "type": "message",
+                "message": {
+                    "role": source,
+                    "content": _text_content(step.get("message")),
+                },
+            }
+        )
+    return rows
+
+
+def _normalize(parsed: Any) -> list[Any]:
+    if isinstance(parsed, list):
+        return parsed
+    if not isinstance(parsed, dict):
+        return []
+
+    schema_version = parsed.get("schema_version")
+    if isinstance(schema_version, str) and schema_version.startswith("ATIF-"):
+        return _atif_to_openclaw(parsed)
+
+    for key in ("transcript", "messages", "chat"):
+        value = parsed.get(key)
+        if isinstance(value, list):
+            return value
+    return [parsed]
+
+
+def _read_transcript_file(path: Path) -> list[Any]:
+    if not path.exists():
+        return []
+    try:
+        raw = path.read_text(encoding="utf-8", errors="ignore")
+    except OSError:
+        return []
+
+    parsed = _safe_json_loads(raw)
+    if parsed is not None:
+        return _normalize(parsed)
+    return _parse_json_lines(raw)
+
+
+def load_transcript(path_str: str = "") -> list[Any]:
+    candidates = ([Path(path_str)] if path_str else []) + list(TRANSCRIPT_CANDIDATES)
+    seen: set[Path] = set()
+    for candidate in candidates:
+        if candidate in seen:
+            continue
+        seen.add(candidate)
+        loaded = _read_transcript_file(candidate)
+        if loaded:
+            return loaded
+    return []
+
+
+def write_openclaw_compat_transcript(
+    transcript: list[Any],
+    path: Path = OPENCLAW_COMPAT_PATH,
+) -> None:
+    """Expose the current Harbor trace at the path used by legacy WCB graders."""
+    if not transcript:
+        return
+    try:
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.write_text(
+            "".join(
+                json.dumps(row, ensure_ascii=False, default=str) + "\n"
+                for row in transcript
+            ),
+            encoding="utf-8",
+        )
+    except OSError:
+        # The transcript is still supplied through grade(transcript=...), so a
+        # non-root verifier can continue even if it cannot write under /root.
+        return

01_Productivity_Flow_task_1_arxiv_digest/environment/.wildclaw/run-warmup.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+marker="/tmp/wildclaw-harbor-warmup.done"
+[[ -f "$marker" ]] && exit 0
+cd /tmp_workspace
+
+bash -lc 'npm install -g agent-browser'
+
+touch "$marker"

01_Productivity_Flow_task_1_arxiv_digest/environment/.wildclaw/skills/agent-browser/SKILL.md

@@ -0,0 +1,206 @@
+---
+name: agent-browser
+description: Headless browser automation CLI optimized for AI agents with accessibility tree snapshots and ref-based element selection
+metadata: {"clawdbot":{"emoji":"🌐","requires":{"commands":["agent-browser"]},"homepage":"https://github.com/vercel-labs/agent-browser"}}
+---
+
+# Agent Browser Skill
+
+Fast browser automation using accessibility tree snapshots with refs for deterministic element selection.
+
+## Why Use This Over Built-in Browser Tool
+
+**Use agent-browser when:**
+- Automating multi-step workflows
+- Need deterministic element selection
+- Performance is critical
+- Working with complex SPAs
+- Need session isolation
+
+**Use built-in browser tool when:**
+- Need screenshots/PDFs for analysis
+- Visual inspection required
+- Browser extension integration needed
+
+## Core Workflow
+
+```bash
+# 1. Navigate and snapshot
+agent-browser open https://example.com
+agent-browser snapshot -i --json
+
+# 2. Parse refs from JSON, then interact
+agent-browser click @e2
+agent-browser fill @e3 "text"
+
+# 3. Re-snapshot after page changes
+agent-browser snapshot -i --json
+```
+
+## Key Commands
+
+### Navigation
+```bash
+agent-browser open <url>
+agent-browser back | forward | reload | close
+```
+
+### Snapshot (Always use -i --json)
+```bash
+agent-browser snapshot -i --json          # Interactive elements, JSON output
+agent-browser snapshot -i -c -d 5 --json  # + compact, depth limit
+agent-browser snapshot -s "#main" -i      # Scope to selector
+```
+
+### Interactions (Ref-based)
+```bash
+agent-browser click @e2
+agent-browser fill @e3 "text"
+agent-browser type @e3 "text"
+agent-browser hover @e4
+agent-browser check @e5 | uncheck @e5
+agent-browser select @e6 "value"
+agent-browser press "Enter"
+agent-browser scroll down 500
+agent-browser drag @e7 @e8
+```
+
+### Get Information
+```bash
+agent-browser get text @e1 --json
+agent-browser get html @e2 --json
+agent-browser get value @e3 --json
+agent-browser get attr @e4 "href" --json
+agent-browser get title --json
+agent-browser get url --json
+agent-browser get count ".item" --json
+```
+
+### Check State
+```bash
+agent-browser is visible @e2 --json
+agent-browser is enabled @e3 --json
+agent-browser is checked @e4 --json
+```
+
+### Wait
+```bash
+agent-browser wait @e2                    # Wait for element
+agent-browser wait 1000                   # Wait ms
+agent-browser wait --text "Welcome"       # Wait for text
+agent-browser wait --url "**/dashboard"   # Wait for URL
+agent-browser wait --load networkidle     # Wait for network
+agent-browser wait --fn "window.ready === true"
+```
+
+### Sessions (Isolated Browsers)
+```bash
+agent-browser --session admin open site.com
+agent-browser --session user open site.com
+agent-browser session list
+# Or via env: AGENT_BROWSER_SESSION=admin agent-browser ...
+```
+
+### State Persistence
+```bash
+agent-browser state save auth.json        # Save cookies/storage
+agent-browser state load auth.json        # Load (skip login)
+```
+
+### Screenshots & PDFs
+```bash
+agent-browser screenshot page.png
+agent-browser screenshot --full page.png
+agent-browser pdf page.pdf
+```
+
+### Network Control
+```bash
+agent-browser network route "**/ads/*" --abort           # Block
+agent-browser network route "**/api/*" --body '{"x":1}'  # Mock
+agent-browser network requests --filter api              # View
+```
+
+### Cookies & Storage
+```bash
+agent-browser cookies                     # Get all
+agent-browser cookies set name value
+agent-browser storage local key           # Get localStorage
+agent-browser storage local set key val
+```
+
+### Tabs & Frames
+```bash
+agent-browser tab new https://example.com
+agent-browser tab 2                       # Switch to tab
+agent-browser frame @e5                   # Switch to iframe
+agent-browser frame main                  # Back to main
+```
+
+## Snapshot Output Format
+
+```json
+{
+  "success": true,
+  "data": {
+    "snapshot": "...",
+    "refs": {
+      "e1": {"role": "heading", "name": "Example Domain"},
+      "e2": {"role": "button", "name": "Submit"},
+      "e3": {"role": "textbox", "name": "Email"}
+    }
+  }
+}
+```
+
+## Best Practices
+
+1. **Always use `-i` flag** - Focus on interactive elements
+2. **Always use `--json`** - Easier to parse
+3. **Wait for stability** - `agent-browser wait --load networkidle`
+4. **Save auth state** - Skip login flows with `state save/load`
+5. **Use sessions** - Isolate different browser contexts
+6. **Use `--headed` for debugging** - See what's happening
+
+## Example: Search and Extract
+
+```bash
+agent-browser open https://www.google.com
+agent-browser snapshot -i --json
+# AI identifies search box @e1
+agent-browser fill @e1 "AI agents"
+agent-browser press Enter
+agent-browser wait --load networkidle
+agent-browser snapshot -i --json
+# AI identifies result refs
+agent-browser get text @e3 --json
+agent-browser get attr @e4 "href" --json
+```
+
+## Example: Multi-Session Testing
+
+```bash
+# Admin session
+agent-browser --session admin open app.com
+agent-browser --session admin state load admin-auth.json
+agent-browser --session admin snapshot -i --json
+
+# User session (simultaneous)
+agent-browser --session user open app.com
+agent-browser --session user state load user-auth.json
+agent-browser --session user snapshot -i --json
+```
+
+## Installation
+
+```bash
+npm install -g agent-browser
+agent-browser install                     # Download Chromium
+agent-browser install --with-deps         # Linux: + system deps
+```
+
+## Credits
+
+Skill created by Yossi Elkrief ([@MaTriXy](https://github.com/MaTriXy))
+
+agent-browser CLI by [Vercel Labs](https://github.com/vercel-labs/agent-browser)

01_Productivity_Flow_task_1_arxiv_digest/environment/.wildclaw/skills/agent-browser/_meta.json

@@ -0,0 +1,6 @@
+{
+  "ownerId": "kn7amrtkn0tjk2r2yxf3hjgp0s7zn6g4",
+  "slug": "agent-browser-clawdbot",
+  "version": "0.1.0",
+  "publishedAt": 1769032854381
+}

01_Productivity_Flow_task_1_arxiv_digest/instruction.md

@@ -0,0 +1,79 @@
+You are an expert in a restricted, non-interactive environment. Solve the task efficiently before the timeout (1200s). Use the available tools and as many iterative steps as needed. Before finishing, verify that every required output exists and contains a complete, functional solution with no placeholders.
+
+I'm a CV researcher and the author of **CapRL**.
+
+Please prepare my daily arxiv paper digest:
+
+1. Fetch cs.CV papers submitted on 2026-02-25 from the arxiv.
+
+2. From all fetched papers, classify every paper into exactly one of the following categories and group them accordingly. Papers that don't fit the first five topic categories must be placed under "Others". In the Classification section, only list paper titles (you may include arXiv IDs). No per-paper explanation or summary is needed.
+   - Multimodal / Vision-Language Models
+   - Medical Image Analysis
+   - Image / Video Generation & Editing
+   - Autonomous Driving / Robotics / Embodied AI
+   - 3D Vision / Reconstruction / Gaussian Splatting
+   - Others
+
+3. For the papers classified under "Multimodal / Vision-Language Models", build a metadata audit table:
+   - arXiv ID
+   - all authors
+   - whether the paper has an appendix / supplementary section
+   - number of main-paper figures
+   - number of appendix figures
+   - number of main-paper tables
+   - number of appendix tables
+   - total number of figures, including appendix figures
+   - total number of tables, including appendix tables
+   - appendix evidence: if the paper has an appendix, give the first appendix / supplementary heading you used as the boundary; otherwise write `None`
+
+   Count figures/tables from the paper content rather than from abstract-page metadata. If you use arXiv HTML, count top-level paper figures/tables instead of subfigure markers such as "(a)" and "(b)". The `Authors` field must cover the complete author list, not a partial subset. The total counts must equal `main + appendix`.
+
+4. Based on my research interests, highlight papers I might find interesting.
+
+5. If any paper benchmarks against CapRL, extract the comparison results.
+
+Save everything to `/tmp_workspace/results/arxiv_digest.md` using exactly the following structure (do not rename these section headings):
+
+```markdown
+# ArXiv Daily Digest - 2026-02-25
+
+### Classification
+#### Multimodal / Vision-Language Models
+- **Paper Title** (arXiv ID)
+
+#### Medical Image Analysis
+- ...
+
+#### Image / Video Generation & Editing
+- ...
+
+#### Autonomous Driving / Robotics / Embodied AI
+- ...
+
+#### 3D Vision / Reconstruction / Gaussian Splatting
+- ...
+
+#### Others
+- ...
+
+### Multimodal Paper Metadata Audit
+For papers in the "Multimodal / Vision-Language Models" category, report the full author list, whether the paper has an appendix, split figure/table counts between main paper and appendix, the totals, and appendix evidence:
+
+| Paper | arXiv ID | Authors | Has Appendix | Main Figures | Appendix Figures | Main Tables | Appendix Tables | Total Figures | Total Tables | Appendix Evidence |
+|------|----------|---------|--------------|--------------|------------------|-------------|-----------------|---------------|--------------|-------------------|
+| ... | ... | ... | Yes / No | ... | ... | ... | ... | ... | ... | ... |
+
+### Personalized Recommendations
+#### Papers of Interest
+Select exactly 1 paper most relevant to my research interests, with a brief reason.
+- **Paper Title** — why it's relevant
+
+#### Benchmark Comparison
+If any paper compares against CapRL, extract the **Prism evaluation** main table in markdown table format, focusing on CapRL and the paper's proposed method:
+
+| MLLM | Benchmark1 | Benchmark2 | ... | Avg. |
+|------|------------|------------|-----|------|
+| CapRL-3B | ... | ... | ... | <avg> |
+| [proposed method] | ... | ... | ... | <avg> |
+| ... |
+```

01_Productivity_Flow_task_1_arxiv_digest/task.toml

@@ -0,0 +1,29 @@
+schema_version = "1.4"
+artifacts = ["/tmp_workspace/results"]
+
+[metadata]
+benchmark = "WildClawBench"
+source_task_id = "01_Productivity_Flow_task_1_arxiv_digest"
+name = "ArXiv Daily Paper Digest"
+category = "01_Productivity_Flow"
+modality = "pure-text"
+
+[agent]
+timeout_sec = 1200.0
+
+[verifier]
+environment_mode = "shared"
+timeout_sec = 600.0
+
+[environment]
+docker_image = "wildclawbench-ubuntu:v1.3"
+workdir = "/tmp_workspace"
+network_mode = "public"
+env = { BRAVE_API_KEY = "${BRAVE_API_KEY:-}", HTTPS_PROXY = "", HTTP_PROXY = "", NO_PROXY = "localhost,127.0.0.1,host.docker.internal", http_proxy = "", https_proxy = "", no_proxy = "localhost,127.0.0.1,host.docker.internal" }
+skills_dir = "/tmp_workspace/.wildclaw/skills"
+
+[environment.healthcheck]
+command = "bash /tmp_workspace/.wildclaw/run-warmup.sh"
+timeout_sec = 1200.0
+interval_sec = 1.0
+retries = 1

01_Productivity_Flow_task_1_arxiv_digest/tests/checks.py

@@ -0,0 +1,535 @@
+def grade(**kwargs) -> dict:
+    """
+    Grade the arxiv digest task.
+
+    Returns:
+        Dict mapping criterion names to scores (0.0 to 1.0)
+    """
+    from pathlib import Path
+    import re
+
+    ALL_CRITERIA = [
+        "classify_multimodal",
+        "classify_medical",
+        "classify_generation",
+        "classify_driving",
+        "classify_3d",
+        "metadata_appendix_flags",
+        "metadata_split_consistency",
+        "metadata_simpleocr",
+        "metadata_exploring_multimodal_lmms",
+        "metadata_nolan",
+        "metadata_weavetime",
+        "metadata_global_local_dual_perception",
+        "metadata_see_it_say_it_sorted",
+        "metadata_dynamicgtr",
+        "metadata_dynamic_multimodal_activation_steering",
+        "metadata_dr_seg",
+        "metadata_cccaption",
+        "interest_count",
+        "interest_selection",
+        "benchmark_charxiv",
+        "benchmark_caprl_avg",
+        "benchmark_cccaption_avg",
+        "benchmark_caprl_infovqa",
+        "classify_score",
+        "metadata_score",
+        "interest_score",
+        "overall_score",
+    ]
+
+    scores = {}
+    workspace = Path("/tmp_workspace/results")
+    digest = workspace / "arxiv_digest.md"
+
+    if not digest.exists() or len(digest.read_text().strip()) < 100:
+        return {k: 0.0 for k in ALL_CRITERIA}
+
+    content = digest.read_text()
+
+    def extract_section(markdown_text: str, heading_pattern: str) -> str:
+        """
+        Extract section content under the first heading that matches heading_pattern.
+        Section ends at the next heading of the same or higher level.
+        """
+        h = re.search(heading_pattern, markdown_text, re.MULTILINE | re.IGNORECASE)
+        if not h:
+            return ""
+
+        heading_line = h.group(0)
+        level_match = re.match(r"^#{1,6}", heading_line)
+        if not level_match:
+            return ""
+        level = len(level_match.group(0))
+
+        rest = markdown_text[h.end():]
+        next_h = re.search(rf"^#{{1,{level}}}\s+", rest, re.MULTILINE)
+        return rest[: next_h.start()] if next_h else rest
+
+    # --- Classification accuracy ---
+    # Strictly scope to the "### Classification" block, then require exact
+    # matches for the category "####" sub-headings.
+    strict_headings = [
+        ("multimodal", "Multimodal / Vision-Language Models"),
+        ("medical", "Medical Image Analysis"),
+        ("generation", "Image / Video Generation & Editing"),
+        ("driving", "Autonomous Driving / Robotics / Embodied AI"),
+        ("3d", "3D Vision / Reconstruction / Gaussian Splatting"),
+    ]
+
+    classification_section = extract_section(content, r"^###\s+Classification\s*$")
+
+    # Map strict headings to their positions
+    strict_heading_map = {}
+    for cat_id, heading_text in strict_headings:
+        m = re.search(
+            rf"^####\s+{re.escape(heading_text)}\s*$",
+            classification_section,
+            re.MULTILINE,
+        )
+        if m:
+            strict_heading_map[m.start()] = (m.start(), m.end(), cat_id)
+
+    # Find ALL #### headings (including "Others") to use as boundaries
+    all_h4 = [(m.start(), m.end()) for m in re.finditer(r"^####\s+", classification_section, re.MULTILINE)]
+    all_h4.sort()
+
+    # Extract section text: from each strict heading to the next #### heading
+    sections = {}
+    for idx, (h_start, h_end) in enumerate(all_h4):
+        if h_start not in strict_heading_map:
+            continue
+        _, s_end, cat_id = strict_heading_map[h_start]
+        next_starts = [s for s, _ in all_h4 if s > h_start]
+        section_end = next_starts[0] if next_starts else len(classification_section)
+        sections[cat_id] = classification_section[s_end:section_end]
+
+    # Ground-truth: checkpoint papers per category (distinctive keywords)
+    ground_truth = {
+        "multimodal": [
+            r"SimpleOCR",
+            r"Exploring Multimodal LMMs",
+            r"NoLan",
+            r"WeaveTime",
+            r"Global.Local Dual Perception",
+            r"See It, Say It, Sorted",
+            r"DynamicGTR",
+            r"Dynamic Multimodal Activation Steering",
+            r"Dr\.\s*Seg",
+            r"CCCaption",
+        ],
+        "medical": [
+            r"(?:Diagnostic Trace|Visual Cognition.guided.*Chest X.Ray)",
+            r"(?:Brain Tumor Segmentation.*Non.Enhancing)",
+            r"SigVLP",
+        ],
+        "generation": [
+            r"SkyReels.?V4",
+            r"MultiAnimate",
+            r"(?:Accelerating Diffusion.*Pipeline|Hybrid Data.Pipeline.*Diffusion)",
+        ],
+        "driving": [
+            r"(?:World Guidance|World Modeling.*Condition Space)",
+            r"LiLo.VLA",
+            r"SEF.MAP",
+        ],
+        "3d": [
+            r"(?:Visual Geometry Priors.*Gaussian|Sparse Gaussian Occupancy)",
+            r"(?:Cryo.EM|Protein.*Cryo)",
+            r"UniHand",
+        ],
+    }
+
+    for cat, papers in ground_truth.items():
+        section_text = sections.get(cat, "")
+        correct = sum(1 for p in papers if re.search(p, section_text, re.IGNORECASE))
+        scores[f"classify_{cat}"] = round(correct / len(papers), 2)
+
+    # --- Multimodal metadata audit ---
+    metadata_section = extract_section(
+        content,
+        r"^###\s+Multimodal Paper Metadata Audit\s*$",
+    )
+    metadata_section_exists = bool(metadata_section.strip())
+
+    metadata_ground_truth = {
+        "metadata_simpleocr": {
+            "paper": r"SimpleOCR",
+            "authors": [
+                r"Yibo\s+Peng",
+                r"Peng\s+Xia",
+                r"Ding\s+Zhong",
+                r"Kaide\s+Zeng",
+                r"Siwei\s+Han",
+                r"Yiyang\s+Zhou",
+                r"Jiaqi\s+Liu",
+                r"Ruiyi\s+Zhang",
+                r"Huaxiu\s+Yao",
+            ],
+            "has_appendix": True,
+            "main_figures": 5,
+            "appendix_figures": 0,
+            "main_tables": 4,
+            "appendix_tables": 4,
+            "total_figures": 5,
+            "total_tables": 8,
+            "appendix_evidence": r"Appendix A Dataset Details",
+        },
+        "metadata_exploring_multimodal_lmms": {
+            "paper": r"Exploring Multimodal LMMs",
+            "authors": [
+                r"Giuseppe\s+Lando",
+                r"Rosario\s+Forte",
+                r"Antonino\s+Furnari",
+            ],
+            "has_appendix": False,
+            "main_figures": 3,
+            "appendix_figures": 0,
+            "main_tables": 5,
+            "appendix_tables": 0,
+            "total_figures": 3,
+            "total_tables": 5,
+            "appendix_evidence": r"None",
+        },
+        "metadata_nolan": {
+            "paper": r"NoLan",
+            "authors": [
+                r"Lingfeng\s+Ren",
+                r"Weihao\s+Yu",
+                r"Runpeng\s+Yu",
+                r"Xinchao\s+Wang",
+            ],
+            "has_appendix": True,
+            "main_figures": 4,
+            "appendix_figures": 2,
+            "main_tables": 5,
+            "appendix_tables": 17,
+            "total_figures": 6,
+            "total_tables": 22,
+            "appendix_evidence": r"Appendix A Appendix",
+        },
+        "metadata_weavetime": {
+            "paper": r"WeaveTime",
+            "authors": [
+                r"Yulin\s+Zhang",
+                r"Cheng\s+Shi",
+                r"Sibei\s+Yang",
+            ],
+            "has_appendix": False,
+            "main_figures": 7,
+            "appendix_figures": 0,
+            "main_tables": 5,
+            "appendix_tables": 0,
+            "total_figures": 7,
+            "total_tables": 5,
+            "appendix_evidence": r"None",
+        },
+        "metadata_global_local_dual_perception": {
+            "paper": r"Global.Local Dual Perception",
+            "authors": [
+                r"Junxin\s+Lu",
+                r"Tengfei\s+Song",
+                r"Zhanglin\s+Wu",
+                r"Pengfei\s+Li",
+                r"Xiaowei\s+Liang",
+                r"Hui\s+Yang",
+                r"Kun\s+Chen",
+                r"Ning\s+Xie",
+                r"Yunfei\s+Lu",
+                r"Jing\s+Zhao",
+                r"Shiliang\s+Sun",
+                r"Daimeng\s+Wei",
+            ],
+            "has_appendix": False,
+            "main_figures": 7,
+            "appendix_figures": 0,
+            "main_tables": 4,
+            "appendix_tables": 0,
+            "total_figures": 7,
+            "total_tables": 4,
+            "appendix_evidence": r"None",
+        },
+        "metadata_see_it_say_it_sorted": {
+            "paper": r"See It, Say It, Sorted",
+            "authors": [
+                r"Yongchang\s+Zhang",
+                r"Oliver\s+Ma",
+                r"Tianyi\s+Liu",
+                r"Guangquan\s+Zhou",
+                r"Yang\s+Chen",
+            ],
+            "has_appendix": False,
+            "main_figures": 5,
+            "appendix_figures": 0,
+            "main_tables": 6,
+            "appendix_tables": 0,
+            "total_figures": 5,
+            "total_tables": 6,
+            "appendix_evidence": r"None",
+        },
+        "metadata_dynamicgtr": {
+            "paper": r"DynamicGTR",
+            "authors": [
+                r"Yanbin\s+Wei",
+                r"Jiangyue\s+Yan",
+                r"Chun\s+Kang",
+                r"Yang\s+Chen",
+                r"Hua\s+Liu",
+                r"James\s+Kwok",
+                r"Yu\s+Zhang",
+            ],
+            "has_appendix": True,
+            "main_figures": 3,
+            "appendix_figures": 1,
+            "main_tables": 8,
+            "appendix_tables": 11,
+            "total_figures": 4,
+            "total_tables": 19,
+            "appendix_evidence": r"A\.?\s*GTR Generation",
+        },
+        "metadata_dynamic_multimodal_activation_steering": {
+            "paper": r"Dynamic Multimodal Activation Steering",
+            "authors": [
+                r"Jianghao\s+Yin",
+                r"Qin\s+Chen",
+                r"Kedi\s+Chen",
+                r"Jie\s+Zhou",
+                r"Xingjiao\s+Wu",
+                r"Liang\s+He",
+            ],
+            "has_appendix": True,
+            "main_figures": 4,
+            "appendix_figures": 2,
+            "main_tables": 5,
+            "appendix_tables": 9,
+            "total_figures": 6,
+            "total_tables": 14,
+            "appendix_evidence": r"Appendix A Appendix",
+        },
+        "metadata_dr_seg": {
+            "paper": r"Dr\.\s*Seg",
+            "authors": [
+                r"Haoxiang\s+Sun",
+                r"Tao\s+Wang",
+                r"Chenwei\s+Tang",
+                r"Li\s+Yuan",
+                r"Jiancheng\s+Lv",
+            ],
+            "has_appendix": True,
+            "main_figures": 7,
+            "appendix_figures": 7,
+            "main_tables": 6,
+            "appendix_tables": 6,
+            "total_figures": 14,
+            "total_tables": 12,
+            "appendix_evidence": r"Appendix A More Experiment Details and Ablations",
+        },
+        "metadata_cccaption": {
+            "paper": r"CCCaption",
+            "authors": [
+                r"Zhijiang\s+Tang",
+                r"Linhua\s+Wang",
+                r"Jiaxin\s+Qi",
+                r"Weihao\s+Jiang",
+                r"Peng\s+Hou",
+                r"Anxiang\s+Zeng",
+                r"Jianqiang\s+Huang",
+            ],
+            "has_appendix": False,
+            "main_figures": 5,
+            "appendix_figures": 0,
+            "main_tables": 5,
+            "appendix_tables": 0,
+            "total_figures": 5,
+            "total_tables": 5,
+            "appendix_evidence": r"None",
+        },
+    }
+
+    appendix_flags_ok = True
+    split_consistency_ok = True
+    if metadata_section_exists:
+        for score_key, spec in metadata_ground_truth.items():
+            row_match = re.search(
+                rf"^\|[^\n]*{spec['paper']}[^\n]*\|\s*$",
+                metadata_section,
+                re.IGNORECASE | re.MULTILINE,
+            )
+            if not row_match:
+                appendix_flags_ok = False
+                split_consistency_ok = False
+                scores[score_key] = 0.0
+                continue
+
+            row_text = row_match.group(0)
+            author_ok = all(re.search(author_pat, row_text, re.IGNORECASE) for author_pat in spec["authors"])
+            appendix_ok = bool(
+                re.search(
+                    r"\|\s*(?:yes|true)\s*\|",
+                    row_text,
+                    re.IGNORECASE,
+                )
+            ) if spec["has_appendix"] else bool(
+                re.search(
+                    r"\|\s*(?:no|false)\s*\|",
+                    row_text,
+                    re.IGNORECASE,
+                )
+            )
+            main_figures_ok = bool(re.search(rf"\|\s*{spec['main_figures']}\s*\|", row_text))
+            appendix_figures_ok = bool(re.search(rf"\|\s*{spec['appendix_figures']}\s*\|", row_text))
+            main_tables_ok = bool(re.search(rf"\|\s*{spec['main_tables']}\s*\|", row_text))
+            appendix_tables_ok = bool(re.search(rf"\|\s*{spec['appendix_tables']}\s*\|", row_text))
+            total_figures_ok = bool(re.search(rf"\|\s*{spec['total_figures']}\s*\|", row_text))
+            total_tables_ok = bool(re.search(rf"\|\s*{spec['total_tables']}\s*\|", row_text))
+            evidence_ok = bool(re.search(spec["appendix_evidence"], row_text, re.IGNORECASE))
+
+            cells = [c.strip() for c in row_text.strip().strip("|").split("|")]
+            if len(cells) >= 10:
+                try:
+                    row_main_figures = int(cells[4])
+                    row_appendix_figures = int(cells[5])
+                    row_main_tables = int(cells[6])
+                    row_appendix_tables = int(cells[7])
+                    row_total_figures = int(cells[8])
+                    row_total_tables = int(cells[9])
+                    split_figures_consistent = row_main_figures + row_appendix_figures == row_total_figures
+                    split_tables_consistent = row_main_tables + row_appendix_tables == row_total_tables
+                except ValueError:
+                    split_figures_consistent = False
+                    split_tables_consistent = False
+            else:
+                split_figures_consistent = False
+                split_tables_consistent = False
+            appendix_flags_ok = appendix_flags_ok and appendix_ok
+            split_consistency_ok = split_consistency_ok and split_figures_consistent and split_tables_consistent
+            scores[score_key] = 1.0 if (
+                author_ok
+                and main_figures_ok
+                and appendix_figures_ok
+                and main_tables_ok
+                and appendix_tables_ok
+                and total_figures_ok
+                and total_tables_ok
+                and evidence_ok
+            ) else 0.0
+    else:
+        for score_key in metadata_ground_truth:
+            scores[score_key] = 0.0
+        appendix_flags_ok = False
+        split_consistency_ok = False
+
+    scores["metadata_appendix_flags"] = 1.0 if (metadata_section_exists and appendix_flags_ok) else 0.0
+    scores["metadata_split_consistency"] = 1.0 if (metadata_section_exists and split_consistency_ok) else 0.0
+
+    # --- Interest selection ---
+    # "### Papers of Interest" must contain exactly one recommended paper, and
+    # that paper should be CCCaption.
+    # We first scope to "## Personalized Recommendations", then search inside it.
+    recommendations_section = extract_section(
+        content,
+        r"^#{2,4}\s+[^\n]*(?:[Pp]ersonali|[Rr]ecommend)[^\n]*$",
+    )
+    interest_section = extract_section(
+        recommendations_section if recommendations_section.strip() else content,
+        r"^#{2,4}\s+[^\n]*[Pp]apers?\s+[Oo]f\s+[Ii]nterest[^\n]*$",
+    )
+    recommended_papers = re.findall(
+        r"(?m)^(?:-\s+\*\*.+?\*\*|\d+\.\s+\*\*.+?\*\*|-\s+.+?—.+|\d+\.\s+.+?—.+)$",
+        interest_section,
+    )
+    scores["interest_count"] = 1.0 if len(recommended_papers) == 1 else 0.0
+    scores["interest_selection"] = (
+        1.0
+        if (
+            scores["interest_count"] == 1.0
+            and re.search(r"CCCaption", interest_section, re.IGNORECASE)
+        )
+        else 0.0
+    )
+
+    # --- Benchmark extraction sub-items ---
+    benchmark_source = recommendations_section if recommendations_section.strip() else content
+    benchmark_section = extract_section(
+        benchmark_source,
+        r"^#{2,4}\s+[^\n]*(?:[Bb]enchmark|[Cc]omparison|[Pp]rism)[^\n]*$",
+    )
+
+    # Sub-item 1: benchmark table contains CharXiv
+    has_charxiv = bool(
+        re.search(r"^\|[^\n]*CharXiv[^\n]*\|\s*$", benchmark_section, re.IGNORECASE | re.MULTILINE)
+    )
+    scores["benchmark_charxiv"] = 1.0 if has_charxiv else 0.0
+
+    # Sub-item 2: CapRL-3B average score is 51.07
+    scores["benchmark_caprl_avg"] = (
+        1.0 if re.search(r"CapRL.{0,10}3B.*?51\.07", benchmark_section, re.IGNORECASE) else 0.0
+    )
+
+    # Sub-item 3: CCCaption-2B average score is 52.80
+    scores["benchmark_cccaption_avg"] = (
+        1.0
+        if re.search(r"CCCaption.{0,10}2B.*?52\.80", benchmark_section, re.IGNORECASE)
+        else 0.0
+    )
+
+    # Sub-item 4: CapRL row has InfoVQA score 55.94
+    has_infovqa_header = bool(re.search(r"\|\s*InfoVQA\s*\|", benchmark_section, re.IGNORECASE))
+    caprl_row_has_5594 = bool(
+        re.search(r"^\|[^\n]*CapRL[^\n]*55\.94[^\n]*\|\s*$", benchmark_section, re.IGNORECASE | re.MULTILINE)
+    )
+    scores["benchmark_caprl_infovqa"] = 1.0 if (has_infovqa_header and caprl_row_has_5594) else 0.0
+
+    classify_keys = [
+        "classify_multimodal",
+        "classify_medical",
+        "classify_generation",
+        "classify_driving",
+        "classify_3d",
+    ]
+    metadata_keys = [
+        "metadata_appendix_flags",
+        "metadata_split_consistency",
+        "metadata_simpleocr",
+        "metadata_exploring_multimodal_lmms",
+        "metadata_nolan",
+        "metadata_weavetime",
+        "metadata_global_local_dual_perception",
+        "metadata_see_it_say_it_sorted",
+        "metadata_dynamicgtr",
+        "metadata_dynamic_multimodal_activation_steering",
+        "metadata_dr_seg",
+        "metadata_cccaption",
+    ]
+    interest_keys = [
+        "interest_count",
+        "interest_selection",
+        "benchmark_charxiv",
+        "benchmark_caprl_avg",
+        "benchmark_cccaption_avg",
+        "benchmark_caprl_infovqa",
+    ]
+
+    scores["classify_score"] = round(
+        sum(scores.get(k, 0.0) for k in classify_keys) / len(classify_keys), 4
+    )
+    scores["metadata_score"] = round(
+        (
+            sum(scores.get(k, 0.0) for k in metadata_keys) / len(metadata_keys)
+            if metadata_section_exists
+            else 0.0
+        ),
+        4,
+    )
+    scores["interest_score"] = round(
+        sum(scores.get(k, 0.0) for k in interest_keys) / len(interest_keys), 4
+    )
+
+    scores["overall_score"] = round(
+        0.3 * scores["classify_score"]
+        + 0.4 * scores["metadata_score"]
+        + 0.3 * scores["interest_score"],
+        4,
+    )
+
+    return scores

01_Productivity_Flow_task_1_arxiv_digest/tests/grader.py

@@ -0,0 +1,76 @@
+"""Execute an embedded WildClawBench grader as a Harbor verifier."""
+
+from __future__ import annotations
+
+import json
+import math
+import runpy
+import traceback
+from pathlib import Path
+from typing import Any
+
+from transcript_loader import load_transcript, write_openclaw_compat_transcript
+
+
+WORKSPACE = "/tmp_workspace"
+TESTS_DIR = Path("/tests")
+VERIFIER_LOG_DIR = Path("/logs/verifier")
+
+
+def _numeric_rewards(scores: dict[str, Any]) -> dict[str, float]:
+    rewards: dict[str, float] = {}
+    for key, value in scores.items():
+        if isinstance(value, bool):
+            rewards[str(key)] = float(value)
+        elif isinstance(value, (int, float)) and math.isfinite(float(value)):
+            rewards[str(key)] = float(value)
+
+    if "overall_score" not in rewards:
+        numeric_values = list(rewards.values())
+        rewards["overall_score"] = (
+            sum(numeric_values) / len(numeric_values) if numeric_values else 0.0
+        )
+    return rewards
+
+
+def _write_json(path: Path, value: Any) -> None:
+    path.write_text(
+        json.dumps(value, indent=2, ensure_ascii=False, default=str) + "\n",
+        encoding="utf-8",
+    )
+
+
+def main() -> int:
+    VERIFIER_LOG_DIR.mkdir(parents=True, exist_ok=True)
+    try:
+        namespace = runpy.run_path(str(TESTS_DIR / "checks.py"))
+        grade = namespace.get("grade")
+        if not callable(grade):
+            raise TypeError("/tests/checks.py must define a callable grade()")
+
+        transcript = load_transcript()
+        write_openclaw_compat_transcript(transcript)
+        scores = grade(
+            transcript=transcript,
+            workspace_path=WORKSPACE,
+        )
+        if not isinstance(scores, dict):
+            raise TypeError(
+                f"WildClawBench grade() returned {type(scores).__name__}, expected dict"
+            )
+
+        _write_json(VERIFIER_LOG_DIR / "wildclaw_score.json", scores)
+        _write_json(VERIFIER_LOG_DIR / "reward.json", _numeric_rewards(scores))
+        return 0
+    except Exception as exc:
+        traceback.print_exc()
+        _write_json(
+            VERIFIER_LOG_DIR / "wildclaw_score.json",
+            {"overall_score": 0.0, "error": str(exc)},
+        )
+        _write_json(VERIFIER_LOG_DIR / "reward.json", {"overall_score": 0.0})
+        return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

01_Productivity_Flow_task_1_arxiv_digest/tests/test.sh

@@ -0,0 +1,11 @@
+#!/usr/bin/env bash
+set -u
+
+mkdir -p /logs/verifier
+
+if [[ -d /tests/gt ]]; then
+  mkdir -p /tmp_workspace/gt
+  cp -a /tests/gt/. /tmp_workspace/gt/
+fi
+
+python3 /tests/grader.py

01_Productivity_Flow_task_1_arxiv_digest/tests/transcript_loader.py

@@ -0,0 +1,191 @@
+"""Load agent transcripts and normalize Harbor ATIF traces for WCB graders."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+from typing import Any
+
+
+OPENCLAW_COMPAT_PATH = Path("/root/.openclaw/agents/main/sessions/chat.jsonl")
+TRANSCRIPT_CANDIDATES = (
+    Path("/logs/agent/trajectory.json"),
+    OPENCLAW_COMPAT_PATH,
+    Path("/claude_code/log/chat.json"),
+)
+
+
+def _safe_json_loads(text: str) -> Any | None:
+    try:
+        return json.loads(text)
+    except Exception:
+        return None
+
+
+def _parse_json_lines(raw: str) -> list[Any]:
+    rows: list[Any] = []
+    for line in raw.splitlines():
+        line = line.strip()
+        if not line:
+            continue
+        parsed = _safe_json_loads(line)
+        if parsed is not None:
+            rows.append(parsed)
+    return rows
+
+
+def _text_content(value: Any) -> str:
+    if isinstance(value, str):
+        return value
+    if not isinstance(value, list):
+        return ""
+
+    texts: list[str] = []
+    for part in value:
+        if not isinstance(part, dict):
+            continue
+        if part.get("type") == "text" and isinstance(part.get("text"), str):
+            texts.append(part["text"])
+    return "\n".join(texts)
+
+
+def _atif_to_openclaw(trajectory: dict[str, Any]) -> list[dict[str, Any]]:
+    """Convert the ATIF fields used by Harbor agents to WCB's message shape."""
+    rows: list[dict[str, Any]] = []
+    steps = trajectory.get("steps")
+    if not isinstance(steps, list):
+        return rows
+
+    for step in steps:
+        if not isinstance(step, dict):
+            continue
+        source = step.get("source")
+        if source not in ("system", "user", "agent"):
+            continue
+
+        if source == "agent":
+            content: list[dict[str, Any]] = []
+            message = _text_content(step.get("message"))
+            if message:
+                content.append({"type": "text", "text": message})
+
+            tool_calls = step.get("tool_calls")
+            if isinstance(tool_calls, list):
+                for tool_call in tool_calls:
+                    if not isinstance(tool_call, dict):
+                        continue
+                    content.append(
+                        {
+                            "type": "toolCall",
+                            "id": str(tool_call.get("tool_call_id", "")),
+                            "name": str(tool_call.get("function_name", "")),
+                            "arguments": tool_call.get("arguments", {}),
+                        }
+                    )
+
+            rows.append(
+                {
+                    "type": "message",
+                    "message": {
+                        "role": "assistant",
+                        "content": content or message,
+                    },
+                }
+            )
+
+            observation = step.get("observation")
+            results = (
+                observation.get("results") if isinstance(observation, dict) else None
+            )
+            if isinstance(results, list):
+                for result in results:
+                    if not isinstance(result, dict):
+                        continue
+                    rows.append(
+                        {
+                            "type": "message",
+                            "message": {
+                                "role": "toolResult",
+                                "toolCallId": result.get("source_call_id"),
+                                "content": result.get("content", ""),
+                            },
+                        }
+                    )
+            continue
+
+        rows.append(
+            {
+                "type": "message",
+                "message": {
+                    "role": source,
+                    "content": _text_content(step.get("message")),
+                },
+            }
+        )
+    return rows
+
+
+def _normalize(parsed: Any) -> list[Any]:
+    if isinstance(parsed, list):
+        return parsed
+    if not isinstance(parsed, dict):
+        return []
+
+    schema_version = parsed.get("schema_version")
+    if isinstance(schema_version, str) and schema_version.startswith("ATIF-"):
+        return _atif_to_openclaw(parsed)
+
+    for key in ("transcript", "messages", "chat"):
+        value = parsed.get(key)
+        if isinstance(value, list):
+            return value
+    return [parsed]
+
+
+def _read_transcript_file(path: Path) -> list[Any]:
+    if not path.exists():
+        return []
+    try:
+        raw = path.read_text(encoding="utf-8", errors="ignore")
+    except OSError:
+        return []
+
+    parsed = _safe_json_loads(raw)
+    if parsed is not None:
+        return _normalize(parsed)
+    return _parse_json_lines(raw)
+
+
+def load_transcript(path_str: str = "") -> list[Any]:
+    candidates = ([Path(path_str)] if path_str else []) + list(TRANSCRIPT_CANDIDATES)
+    seen: set[Path] = set()
+    for candidate in candidates:
+        if candidate in seen:
+            continue
+        seen.add(candidate)
+        loaded = _read_transcript_file(candidate)
+        if loaded:
+            return loaded
+    return []
+
+
+def write_openclaw_compat_transcript(
+    transcript: list[Any],
+    path: Path = OPENCLAW_COMPAT_PATH,
+) -> None:
+    """Expose the current Harbor trace at the path used by legacy WCB graders."""
+    if not transcript:
+        return
+    try:
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.write_text(
+            "".join(
+                json.dumps(row, ensure_ascii=False, default=str) + "\n"
+                for row in transcript
+            ),
+            encoding="utf-8",
+        )
+    except OSError:
+        # The transcript is still supplied through grade(transcript=...), so a
+        # non-root verifier can continue even if it cannot write under /root.
+        return

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/run-warmup.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+marker="/tmp/wildclaw-harbor-warmup.done"
+[[ -f "$marker" ]] && exit 0
+cd /tmp_workspace
+
+bash -lc 'npm install -g agent-browser'
+
+touch "$marker"

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/agent-browser/SKILL.md

@@ -0,0 +1,206 @@
+---
+name: agent-browser
+description: Headless browser automation CLI optimized for AI agents with accessibility tree snapshots and ref-based element selection
+metadata: {"clawdbot":{"emoji":"🌐","requires":{"commands":["agent-browser"]},"homepage":"https://github.com/vercel-labs/agent-browser"}}
+---
+
+# Agent Browser Skill
+
+Fast browser automation using accessibility tree snapshots with refs for deterministic element selection.
+
+## Why Use This Over Built-in Browser Tool
+
+**Use agent-browser when:**
+- Automating multi-step workflows
+- Need deterministic element selection
+- Performance is critical
+- Working with complex SPAs
+- Need session isolation
+
+**Use built-in browser tool when:**
+- Need screenshots/PDFs for analysis
+- Visual inspection required
+- Browser extension integration needed
+
+## Core Workflow
+
+```bash
+# 1. Navigate and snapshot
+agent-browser open https://example.com
+agent-browser snapshot -i --json
+
+# 2. Parse refs from JSON, then interact
+agent-browser click @e2
+agent-browser fill @e3 "text"
+
+# 3. Re-snapshot after page changes
+agent-browser snapshot -i --json
+```
+
+## Key Commands
+
+### Navigation
+```bash
+agent-browser open <url>
+agent-browser back | forward | reload | close
+```
+
+### Snapshot (Always use -i --json)
+```bash
+agent-browser snapshot -i --json          # Interactive elements, JSON output
+agent-browser snapshot -i -c -d 5 --json  # + compact, depth limit
+agent-browser snapshot -s "#main" -i      # Scope to selector
+```
+
+### Interactions (Ref-based)
+```bash
+agent-browser click @e2
+agent-browser fill @e3 "text"
+agent-browser type @e3 "text"
+agent-browser hover @e4
+agent-browser check @e5 | uncheck @e5
+agent-browser select @e6 "value"
+agent-browser press "Enter"
+agent-browser scroll down 500
+agent-browser drag @e7 @e8
+```
+
+### Get Information
+```bash
+agent-browser get text @e1 --json
+agent-browser get html @e2 --json
+agent-browser get value @e3 --json
+agent-browser get attr @e4 "href" --json
+agent-browser get title --json
+agent-browser get url --json
+agent-browser get count ".item" --json
+```
+
+### Check State
+```bash
+agent-browser is visible @e2 --json
+agent-browser is enabled @e3 --json
+agent-browser is checked @e4 --json
+```
+
+### Wait
+```bash
+agent-browser wait @e2                    # Wait for element
+agent-browser wait 1000                   # Wait ms
+agent-browser wait --text "Welcome"       # Wait for text
+agent-browser wait --url "**/dashboard"   # Wait for URL
+agent-browser wait --load networkidle     # Wait for network
+agent-browser wait --fn "window.ready === true"
+```
+
+### Sessions (Isolated Browsers)
+```bash
+agent-browser --session admin open site.com
+agent-browser --session user open site.com
+agent-browser session list
+# Or via env: AGENT_BROWSER_SESSION=admin agent-browser ...
+```
+
+### State Persistence
+```bash
+agent-browser state save auth.json        # Save cookies/storage
+agent-browser state load auth.json        # Load (skip login)
+```
+
+### Screenshots & PDFs
+```bash
+agent-browser screenshot page.png
+agent-browser screenshot --full page.png
+agent-browser pdf page.pdf
+```
+
+### Network Control
+```bash
+agent-browser network route "**/ads/*" --abort           # Block
+agent-browser network route "**/api/*" --body '{"x":1}'  # Mock
+agent-browser network requests --filter api              # View
+```
+
+### Cookies & Storage
+```bash
+agent-browser cookies                     # Get all
+agent-browser cookies set name value
+agent-browser storage local key           # Get localStorage
+agent-browser storage local set key val
+```
+
+### Tabs & Frames
+```bash
+agent-browser tab new https://example.com
+agent-browser tab 2                       # Switch to tab
+agent-browser frame @e5                   # Switch to iframe
+agent-browser frame main                  # Back to main
+```
+
+## Snapshot Output Format
+
+```json
+{
+  "success": true,
+  "data": {
+    "snapshot": "...",
+    "refs": {
+      "e1": {"role": "heading", "name": "Example Domain"},
+      "e2": {"role": "button", "name": "Submit"},
+      "e3": {"role": "textbox", "name": "Email"}
+    }
+  }
+}
+```
+
+## Best Practices
+
+1. **Always use `-i` flag** - Focus on interactive elements
+2. **Always use `--json`** - Easier to parse
+3. **Wait for stability** - `agent-browser wait --load networkidle`
+4. **Save auth state** - Skip login flows with `state save/load`
+5. **Use sessions** - Isolate different browser contexts
+6. **Use `--headed` for debugging** - See what's happening
+
+## Example: Search and Extract
+
+```bash
+agent-browser open https://www.google.com
+agent-browser snapshot -i --json
+# AI identifies search box @e1
+agent-browser fill @e1 "AI agents"
+agent-browser press Enter
+agent-browser wait --load networkidle
+agent-browser snapshot -i --json
+# AI identifies result refs
+agent-browser get text @e3 --json
+agent-browser get attr @e4 "href" --json
+```
+
+## Example: Multi-Session Testing
+
+```bash
+# Admin session
+agent-browser --session admin open app.com
+agent-browser --session admin state load admin-auth.json
+agent-browser --session admin snapshot -i --json
+
+# User session (simultaneous)
+agent-browser --session user open app.com
+agent-browser --session user state load user-auth.json
+agent-browser --session user snapshot -i --json
+```
+
+## Installation
+
+```bash
+npm install -g agent-browser
+agent-browser install                     # Download Chromium
+agent-browser install --with-deps         # Linux: + system deps
+```
+
+## Credits
+
+Skill created by Yossi Elkrief ([@MaTriXy](https://github.com/MaTriXy))
+
+agent-browser CLI by [Vercel Labs](https://github.com/vercel-labs/agent-browser)

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/agent-browser/_meta.json

@@ -0,0 +1,6 @@
+{
+  "ownerId": "kn7amrtkn0tjk2r2yxf3hjgp0s7zn6g4",
+  "slug": "agent-browser-clawdbot",
+  "version": "0.1.0",
+  "publishedAt": 1769032854381
+}

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/.learnings/ERRORS.md

@@ -0,0 +1,5 @@
+# Errors Log
+
+Command failures, exceptions, and unexpected behaviors.
+
+---

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/.learnings/FEATURE_REQUESTS.md

@@ -0,0 +1,5 @@
+# Feature Requests
+
+Capabilities requested by user that don't currently exist.
+
+---

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/.learnings/LEARNINGS.md

@@ -0,0 +1,5 @@
+# Learnings Log
+
+Captured learnings, corrections, and discoveries. Review before major tasks.
+
+---

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/SKILL.md

@@ -0,0 +1,647 @@
+---
+name: self-improvement
+description: "Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Claude ('No, that's wrong...', 'Actually...'), (3) User requests a capability that doesn't exist, (4) An external API or tool fails, (5) Claude realizes its knowledge is outdated or incorrect, (6) A better approach is discovered for a recurring task. Also review learnings before major tasks."
+metadata:
+---
+
+# Self-Improvement Skill
+
+Log learnings and errors to markdown files for continuous improvement. Coding agents can later process these into fixes, and important learnings get promoted to project memory.
+
+## Quick Reference
+
+| Situation | Action |
+|-----------|--------|
+| Command/operation fails | Log to `.learnings/ERRORS.md` |
+| User corrects you | Log to `.learnings/LEARNINGS.md` with category `correction` |
+| User wants missing feature | Log to `.learnings/FEATURE_REQUESTS.md` |
+| API/external tool fails | Log to `.learnings/ERRORS.md` with integration details |
+| Knowledge was outdated | Log to `.learnings/LEARNINGS.md` with category `knowledge_gap` |
+| Found better approach | Log to `.learnings/LEARNINGS.md` with category `best_practice` |
+| Simplify/Harden recurring patterns | Log/update `.learnings/LEARNINGS.md` with `Source: simplify-and-harden` and a stable `Pattern-Key` |
+| Similar to existing entry | Link with `**See Also**`, consider priority bump |
+| Broadly applicable learning | Promote to `CLAUDE.md`, `AGENTS.md`, and/or `.github/copilot-instructions.md` |
+| Workflow improvements | Promote to `AGENTS.md` (OpenClaw workspace) |
+| Tool gotchas | Promote to `TOOLS.md` (OpenClaw workspace) |
+| Behavioral patterns | Promote to `SOUL.md` (OpenClaw workspace) |
+
+## OpenClaw Setup (Recommended)
+
+OpenClaw is the primary platform for this skill. It uses workspace-based prompt injection with automatic skill loading.
+
+### Installation
+
+**Via ClawdHub (recommended):**
+```bash
+clawdhub install self-improving-agent
+```
+
+**Manual:**
+```bash
+git clone https://github.com/peterskoett/self-improving-agent.git ~/.openclaw/skills/self-improving-agent
+```
+
+Remade for openclaw from original repo : https://github.com/pskoett/pskoett-ai-skills - https://github.com/pskoett/pskoett-ai-skills/tree/main/skills/self-improvement
+
+### Workspace Structure
+
+OpenClaw injects these files into every session:
+
+```
+~/.openclaw/workspace/
+├── AGENTS.md          # Multi-agent workflows, delegation patterns
+├── SOUL.md            # Behavioral guidelines, personality, principles
+├── TOOLS.md           # Tool capabilities, integration gotchas
+├── MEMORY.md          # Long-term memory (main session only)
+├── memory/            # Daily memory files
+│   └── YYYY-MM-DD.md
+└── .learnings/        # This skill's log files
+    ├── LEARNINGS.md
+    ├── ERRORS.md
+    └── FEATURE_REQUESTS.md
+```
+
+### Create Learning Files
+
+```bash
+mkdir -p ~/.openclaw/workspace/.learnings
+```
+
+Then create the log files (or copy from `assets/`):
+- `LEARNINGS.md` — corrections, knowledge gaps, best practices
+- `ERRORS.md` — command failures, exceptions
+- `FEATURE_REQUESTS.md` — user-requested capabilities
+
+### Promotion Targets
+
+When learnings prove broadly applicable, promote them to workspace files:
+
+| Learning Type | Promote To | Example |
+|---------------|------------|---------|
+| Behavioral patterns | `SOUL.md` | "Be concise, avoid disclaimers" |
+| Workflow improvements | `AGENTS.md` | "Spawn sub-agents for long tasks" |
+| Tool gotchas | `TOOLS.md` | "Git push needs auth configured first" |
+
+### Inter-Session Communication
+
+OpenClaw provides tools to share learnings across sessions:
+
+- **sessions_list** — View active/recent sessions
+- **sessions_history** — Read another session's transcript  
+- **sessions_send** — Send a learning to another session
+- **sessions_spawn** — Spawn a sub-agent for background work
+
+### Optional: Enable Hook
+
+For automatic reminders at session start:
+
+```bash
+# Copy hook to OpenClaw hooks directory
+cp -r hooks/openclaw ~/.openclaw/hooks/self-improvement
+
+# Enable it
+openclaw hooks enable self-improvement
+```
+
+See `references/openclaw-integration.md` for complete details.
+
+---
+
+## Generic Setup (Other Agents)
+
+For Claude Code, Codex, Copilot, or other agents, create `.learnings/` in your project:
+
+```bash
+mkdir -p .learnings
+```
+
+Copy templates from `assets/` or create files with headers.
+
+### Add reference to agent files AGENTS.md, CLAUDE.md, or .github/copilot-instructions.md to remind yourself to log learnings. (this is an alternative to hook-based reminders)
+
+#### Self-Improvement Workflow
+
+When errors or corrections occur:
+1. Log to `.learnings/ERRORS.md`, `LEARNINGS.md`, or `FEATURE_REQUESTS.md`
+2. Review and promote broadly applicable learnings to:
+   - `CLAUDE.md` - project facts and conventions
+   - `AGENTS.md` - workflows and automation
+   - `.github/copilot-instructions.md` - Copilot context
+
+## Logging Format
+
+### Learning Entry
+
+Append to `.learnings/LEARNINGS.md`:
+
+```markdown
+## [LRN-YYYYMMDD-XXX] category
+
+**Logged**: ISO-8601 timestamp
+**Priority**: low | medium | high | critical
+**Status**: pending
+**Area**: frontend | backend | infra | tests | docs | config
+
+### Summary
+One-line description of what was learned
+
+### Details
+Full context: what happened, what was wrong, what's correct
+
+### Suggested Action
+Specific fix or improvement to make
+
+### Metadata
+- Source: conversation | error | user_feedback
+- Related Files: path/to/file.ext
+- Tags: tag1, tag2
+- See Also: LRN-20250110-001 (if related to existing entry)
+- Pattern-Key: simplify.dead_code | harden.input_validation (optional, for recurring-pattern tracking)
+- Recurrence-Count: 1 (optional)
+- First-Seen: 2025-01-15 (optional)
+- Last-Seen: 2025-01-15 (optional)
+
+---
+```
+
+### Error Entry
+
+Append to `.learnings/ERRORS.md`:
+
+```markdown
+## [ERR-YYYYMMDD-XXX] skill_or_command_name
+
+**Logged**: ISO-8601 timestamp
+**Priority**: high
+**Status**: pending
+**Area**: frontend | backend | infra | tests | docs | config
+
+### Summary
+Brief description of what failed
+
+### Error
+```
+Actual error message or output
+```
+
+### Context
+- Command/operation attempted
+- Input or parameters used
+- Environment details if relevant
+
+### Suggested Fix
+If identifiable, what might resolve this
+
+### Metadata
+- Reproducible: yes | no | unknown
+- Related Files: path/to/file.ext
+- See Also: ERR-20250110-001 (if recurring)
+
+---
+```
+
+### Feature Request Entry
+
+Append to `.learnings/FEATURE_REQUESTS.md`:
+
+```markdown
+## [FEAT-YYYYMMDD-XXX] capability_name
+
+**Logged**: ISO-8601 timestamp
+**Priority**: medium
+**Status**: pending
+**Area**: frontend | backend | infra | tests | docs | config
+
+### Requested Capability
+What the user wanted to do
+
+### User Context
+Why they needed it, what problem they're solving
+
+### Complexity Estimate
+simple | medium | complex
+
+### Suggested Implementation
+How this could be built, what it might extend
+
+### Metadata
+- Frequency: first_time | recurring
+- Related Features: existing_feature_name
+
+---
+```
+
+## ID Generation
+
+Format: `TYPE-YYYYMMDD-XXX`
+- TYPE: `LRN` (learning), `ERR` (error), `FEAT` (feature)
+- YYYYMMDD: Current date
+- XXX: Sequential number or random 3 chars (e.g., `001`, `A7B`)
+
+Examples: `LRN-20250115-001`, `ERR-20250115-A3F`, `FEAT-20250115-002`
+
+## Resolving Entries
+
+When an issue is fixed, update the entry:
+
+1. Change `**Status**: pending` → `**Status**: resolved`
+2. Add resolution block after Metadata:
+
+```markdown
+### Resolution
+- **Resolved**: 2025-01-16T09:00:00Z
+- **Commit/PR**: abc123 or #42
+- **Notes**: Brief description of what was done
+```
+
+Other status values:
+- `in_progress` - Actively being worked on
+- `wont_fix` - Decided not to address (add reason in Resolution notes)
+- `promoted` - Elevated to CLAUDE.md, AGENTS.md, or .github/copilot-instructions.md
+
+## Promoting to Project Memory
+
+When a learning is broadly applicable (not a one-off fix), promote it to permanent project memory.
+
+### When to Promote
+
+- Learning applies across multiple files/features
+- Knowledge any contributor (human or AI) should know
+- Prevents recurring mistakes
+- Documents project-specific conventions
+
+### Promotion Targets
+
+| Target | What Belongs There |
+|--------|-------------------|
+| `CLAUDE.md` | Project facts, conventions, gotchas for all Claude interactions |
+| `AGENTS.md` | Agent-specific workflows, tool usage patterns, automation rules |
+| `.github/copilot-instructions.md` | Project context and conventions for GitHub Copilot |
+| `SOUL.md` | Behavioral guidelines, communication style, principles (OpenClaw workspace) |
+| `TOOLS.md` | Tool capabilities, usage patterns, integration gotchas (OpenClaw workspace) |
+
+### How to Promote
+
+1. **Distill** the learning into a concise rule or fact
+2. **Add** to appropriate section in target file (create file if needed)
+3. **Update** original entry:
+   - Change `**Status**: pending` → `**Status**: promoted`
+   - Add `**Promoted**: CLAUDE.md`, `AGENTS.md`, or `.github/copilot-instructions.md`
+
+### Promotion Examples
+
+**Learning** (verbose):
+> Project uses pnpm workspaces. Attempted `npm install` but failed. 
+> Lock file is `pnpm-lock.yaml`. Must use `pnpm install`.
+
+**In CLAUDE.md** (concise):
+```markdown
+## Build & Dependencies
+- Package manager: pnpm (not npm) - use `pnpm install`
+```
+
+**Learning** (verbose):
+> When modifying API endpoints, must regenerate TypeScript client.
+> Forgetting this causes type mismatches at runtime.
+
+**In AGENTS.md** (actionable):
+```markdown
+## After API Changes
+1. Regenerate client: `pnpm run generate:api`
+2. Check for type errors: `pnpm tsc --noEmit`
+```
+
+## Recurring Pattern Detection
+
+If logging something similar to an existing entry:
+
+1. **Search first**: `grep -r "keyword" .learnings/`
+2. **Link entries**: Add `**See Also**: ERR-20250110-001` in Metadata
+3. **Bump priority** if issue keeps recurring
+4. **Consider systemic fix**: Recurring issues often indicate:
+   - Missing documentation (→ promote to CLAUDE.md or .github/copilot-instructions.md)
+   - Missing automation (→ add to AGENTS.md)
+   - Architectural problem (→ create tech debt ticket)
+
+## Simplify & Harden Feed
+
+Use this workflow to ingest recurring patterns from the `simplify-and-harden`
+skill and turn them into durable prompt guidance.
+
+### Ingestion Workflow
+
+1. Read `simplify_and_harden.learning_loop.candidates` from the task summary.
+2. For each candidate, use `pattern_key` as the stable dedupe key.
+3. Search `.learnings/LEARNINGS.md` for an existing entry with that key:
+   - `grep -n "Pattern-Key: <pattern_key>" .learnings/LEARNINGS.md`
+4. If found:
+   - Increment `Recurrence-Count`
+   - Update `Last-Seen`
+   - Add `See Also` links to related entries/tasks
+5. If not found:
+   - Create a new `LRN-...` entry
+   - Set `Source: simplify-and-harden`
+   - Set `Pattern-Key`, `Recurrence-Count: 1`, and `First-Seen`/`Last-Seen`
+
+### Promotion Rule (System Prompt Feedback)
+
+Promote recurring patterns into agent context/system prompt files when all are true:
+
+- `Recurrence-Count >= 3`
+- Seen across at least 2 distinct tasks
+- Occurred within a 30-day window
+
+Promotion targets:
+- `CLAUDE.md`
+- `AGENTS.md`
+- `.github/copilot-instructions.md`
+- `SOUL.md` / `TOOLS.md` for OpenClaw workspace-level guidance when applicable
+
+Write promoted rules as short prevention rules (what to do before/while coding),
+not long incident write-ups.
+
+## Periodic Review
+
+Review `.learnings/` at natural breakpoints:
+
+### When to Review
+- Before starting a new major task
+- After completing a feature
+- When working in an area with past learnings
+- Weekly during active development
+
+### Quick Status Check
+```bash
+# Count pending items
+grep -h "Status\*\*: pending" .learnings/*.md | wc -l
+
+# List pending high-priority items
+grep -B5 "Priority\*\*: high" .learnings/*.md | grep "^## \["
+
+# Find learnings for a specific area
+grep -l "Area\*\*: backend" .learnings/*.md
+```
+
+### Review Actions
+- Resolve fixed items
+- Promote applicable learnings
+- Link related entries
+- Escalate recurring issues
+
+## Detection Triggers
+
+Automatically log when you notice:
+
+**Corrections** (→ learning with `correction` category):
+- "No, that's not right..."
+- "Actually, it should be..."
+- "You're wrong about..."
+- "That's outdated..."
+
+**Feature Requests** (→ feature request):
+- "Can you also..."
+- "I wish you could..."
+- "Is there a way to..."
+- "Why can't you..."
+
+**Knowledge Gaps** (→ learning with `knowledge_gap` category):
+- User provides information you didn't know
+- Documentation you referenced is outdated
+- API behavior differs from your understanding
+
+**Errors** (→ error entry):
+- Command returns non-zero exit code
+- Exception or stack trace
+- Unexpected output or behavior
+- Timeout or connection failure
+
+## Priority Guidelines
+
+| Priority | When to Use |
+|----------|-------------|
+| `critical` | Blocks core functionality, data loss risk, security issue |
+| `high` | Significant impact, affects common workflows, recurring issue |
+| `medium` | Moderate impact, workaround exists |
+| `low` | Minor inconvenience, edge case, nice-to-have |
+
+## Area Tags
+
+Use to filter learnings by codebase region:
+
+| Area | Scope |
+|------|-------|
+| `frontend` | UI, components, client-side code |
+| `backend` | API, services, server-side code |
+| `infra` | CI/CD, deployment, Docker, cloud |
+| `tests` | Test files, testing utilities, coverage |
+| `docs` | Documentation, comments, READMEs |
+| `config` | Configuration files, environment, settings |
+
+## Best Practices
+
+1. **Log immediately** - context is freshest right after the issue
+2. **Be specific** - future agents need to understand quickly
+3. **Include reproduction steps** - especially for errors
+4. **Link related files** - makes fixes easier
+5. **Suggest concrete fixes** - not just "investigate"
+6. **Use consistent categories** - enables filtering
+7. **Promote aggressively** - if in doubt, add to CLAUDE.md or .github/copilot-instructions.md
+8. **Review regularly** - stale learnings lose value
+
+## Gitignore Options
+
+**Keep learnings local** (per-developer):
+```gitignore
+.learnings/
+```
+
+**Track learnings in repo** (team-wide):
+Don't add to .gitignore - learnings become shared knowledge.
+
+**Hybrid** (track templates, ignore entries):
+```gitignore
+.learnings/*.md
+!.learnings/.gitkeep
+```
+
+## Hook Integration
+
+Enable automatic reminders through agent hooks. This is **opt-in** - you must explicitly configure hooks.
+
+### Quick Setup (Claude Code / Codex)
+
+Create `.claude/settings.json` in your project:
+
+```json
+{
+  "hooks": {
+    "UserPromptSubmit": [{
+      "matcher": "",
+      "hooks": [{
+        "type": "command",
+        "command": "./skills/self-improvement/scripts/activator.sh"
+      }]
+    }]
+  }
+}
+```
+
+This injects a learning evaluation reminder after each prompt (~50-100 tokens overhead).
+
+### Full Setup (With Error Detection)
+
+```json
+{
+  "hooks": {
+    "UserPromptSubmit": [{
+      "matcher": "",
+      "hooks": [{
+        "type": "command",
+        "command": "./skills/self-improvement/scripts/activator.sh"
+      }]
+    }],
+    "PostToolUse": [{
+      "matcher": "Bash",
+      "hooks": [{
+        "type": "command",
+        "command": "./skills/self-improvement/scripts/error-detector.sh"
+      }]
+    }]
+  }
+}
+```
+
+### Available Hook Scripts
+
+| Script | Hook Type | Purpose |
+|--------|-----------|---------|
+| `scripts/activator.sh` | UserPromptSubmit | Reminds to evaluate learnings after tasks |
+| `scripts/error-detector.sh` | PostToolUse (Bash) | Triggers on command errors |
+
+See `references/hooks-setup.md` for detailed configuration and troubleshooting.
+
+## Automatic Skill Extraction
+
+When a learning is valuable enough to become a reusable skill, extract it using the provided helper.
+
+### Skill Extraction Criteria
+
+A learning qualifies for skill extraction when ANY of these apply:
+
+| Criterion | Description |
+|-----------|-------------|
+| **Recurring** | Has `See Also` links to 2+ similar issues |
+| **Verified** | Status is `resolved` with working fix |
+| **Non-obvious** | Required actual debugging/investigation to discover |
+| **Broadly applicable** | Not project-specific; useful across codebases |
+| **User-flagged** | User says "save this as a skill" or similar |
+
+### Extraction Workflow
+
+1. **Identify candidate**: Learning meets extraction criteria
+2. **Run helper** (or create manually):
+   ```bash
+   ./skills/self-improvement/scripts/extract-skill.sh skill-name --dry-run
+   ./skills/self-improvement/scripts/extract-skill.sh skill-name
+   ```
+3. **Customize SKILL.md**: Fill in template with learning content
+4. **Update learning**: Set status to `promoted_to_skill`, add `Skill-Path`
+5. **Verify**: Read skill in fresh session to ensure it's self-contained
+
+### Manual Extraction
+
+If you prefer manual creation:
+
+1. Create `skills/<skill-name>/SKILL.md`
+2. Use template from `assets/SKILL-TEMPLATE.md`
+3. Follow [Agent Skills spec](https://agentskills.io/specification):
+   - YAML frontmatter with `name` and `description`
+   - Name must match folder name
+   - No README.md inside skill folder
+
+### Extraction Detection Triggers
+
+Watch for these signals that a learning should become a skill:
+
+**In conversation:**
+- "Save this as a skill"
+- "I keep running into this"
+- "This would be useful for other projects"
+- "Remember this pattern"
+
+**In learning entries:**
+- Multiple `See Also` links (recurring issue)
+- High priority + resolved status
+- Category: `best_practice` with broad applicability
+- User feedback praising the solution
+
+### Skill Quality Gates
+
+Before extraction, verify:
+
+- [ ] Solution is tested and working
+- [ ] Description is clear without original context
+- [ ] Code examples are self-contained
+- [ ] No project-specific hardcoded values
+- [ ] Follows skill naming conventions (lowercase, hyphens)
+
+## Multi-Agent Support
+
+This skill works across different AI coding agents with agent-specific activation.
+
+### Claude Code
+
+**Activation**: Hooks (UserPromptSubmit, PostToolUse)
+**Setup**: `.claude/settings.json` with hook configuration
+**Detection**: Automatic via hook scripts
+
+### Codex CLI
+
+**Activation**: Hooks (same pattern as Claude Code)
+**Setup**: `.codex/settings.json` with hook configuration
+**Detection**: Automatic via hook scripts
+
+### GitHub Copilot
+
+**Activation**: Manual (no hook support)
+**Setup**: Add to `.github/copilot-instructions.md`:
+
+```markdown
+## Self-Improvement
+
+After solving non-obvious issues, consider logging to `.learnings/`:
+1. Use format from self-improvement skill
+2. Link related entries with See Also
+3. Promote high-value learnings to skills
+
+Ask in chat: "Should I log this as a learning?"
+```
+
+**Detection**: Manual review at session end
+
+### OpenClaw
+
+**Activation**: Workspace injection + inter-agent messaging
+**Setup**: See "OpenClaw Setup" section above
+**Detection**: Via session tools and workspace files
+
+### Agent-Agnostic Guidance
+
+Regardless of agent, apply self-improvement when you:
+
+1. **Discover something non-obvious** - solution wasn't immediate
+2. **Correct yourself** - initial approach was wrong
+3. **Learn project conventions** - discovered undocumented patterns
+4. **Hit unexpected errors** - especially if diagnosis was difficult
+5. **Find better approaches** - improved on your original solution
+
+### Copilot Chat Integration
+
+For Copilot users, add this to your prompts when relevant:
+
+> After completing this task, evaluate if any learnings should be logged to `.learnings/` using the self-improvement skill format.
+
+Or use quick prompts:
+- "Log this to learnings"
+- "Create a skill from this solution"
+- "Check .learnings/ for related issues"

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/_meta.json

@@ -0,0 +1,6 @@
+{
+  "ownerId": "kn70cjr952qdec1nx70zs6wefn7ynq2t",
+  "slug": "self-improving-agent",
+  "version": "3.0.5",
+  "publishedAt": 1773760428300
+}

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/assets/LEARNINGS.md

@@ -0,0 +1,45 @@
+# Learnings
+
+Corrections, insights, and knowledge gaps captured during development.
+
+**Categories**: correction | insight | knowledge_gap | best_practice
+**Areas**: frontend | backend | infra | tests | docs | config
+**Statuses**: pending | in_progress | resolved | wont_fix | promoted | promoted_to_skill
+
+## Status Definitions
+
+| Status | Meaning |
+|--------|---------|
+| `pending` | Not yet addressed |
+| `in_progress` | Actively being worked on |
+| `resolved` | Issue fixed or knowledge integrated |
+| `wont_fix` | Decided not to address (reason in Resolution) |
+| `promoted` | Elevated to CLAUDE.md, AGENTS.md, or copilot-instructions.md |
+| `promoted_to_skill` | Extracted as a reusable skill |
+
+## Skill Extraction Fields
+
+When a learning is promoted to a skill, add these fields:
+
+```markdown
+**Status**: promoted_to_skill
+**Skill-Path**: skills/skill-name
+```
+
+Example:
+```markdown
+## [LRN-20250115-001] best_practice
+
+**Logged**: 2025-01-15T10:00:00Z
+**Priority**: high
+**Status**: promoted_to_skill
+**Skill-Path**: skills/docker-m1-fixes
+**Area**: infra
+
+### Summary
+Docker build fails on Apple Silicon due to platform mismatch
+...
+```
+
+---
+

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/assets/SKILL-TEMPLATE.md

@@ -0,0 +1,177 @@
+# Skill Template
+
+Template for creating skills extracted from learnings. Copy and customize.
+
+---
+
+## SKILL.md Template
+
+```markdown
+---
+name: skill-name-here
+description: "Concise description of when and why to use this skill. Include trigger conditions."
+---
+
+# Skill Name
+
+Brief introduction explaining the problem this skill solves and its origin.
+
+## Quick Reference
+
+| Situation | Action |
+|-----------|--------|
+| [Trigger 1] | [Action 1] |
+| [Trigger 2] | [Action 2] |
+
+## Background
+
+Why this knowledge matters. What problems it prevents. Context from the original learning.
+
+## Solution
+
+### Step-by-Step
+
+1. First step with code or command
+2. Second step
+3. Verification step
+
+### Code Example
+
+\`\`\`language
+// Example code demonstrating the solution
+\`\`\`
+
+## Common Variations
+
+- **Variation A**: Description and how to handle
+- **Variation B**: Description and how to handle
+
+## Gotchas
+
+- Warning or common mistake #1
+- Warning or common mistake #2
+
+## Related
+
+- Link to related documentation
+- Link to related skill
+
+## Source
+
+Extracted from learning entry.
+- **Learning ID**: LRN-YYYYMMDD-XXX
+- **Original Category**: correction | insight | knowledge_gap | best_practice
+- **Extraction Date**: YYYY-MM-DD
+```
+
+---
+
+## Minimal Template
+
+For simple skills that don't need all sections:
+
+```markdown
+---
+name: skill-name-here
+description: "What this skill does and when to use it."
+---
+
+# Skill Name
+
+[Problem statement in one sentence]
+
+## Solution
+
+[Direct solution with code/commands]
+
+## Source
+
+- Learning ID: LRN-YYYYMMDD-XXX
+```
+
+---
+
+## Template with Scripts
+
+For skills that include executable helpers:
+
+```markdown
+---
+name: skill-name-here
+description: "What this skill does and when to use it."
+---
+
+# Skill Name
+
+[Introduction]
+
+## Quick Reference
+
+| Command | Purpose |
+|---------|---------|
+| `./scripts/helper.sh` | [What it does] |
+| `./scripts/validate.sh` | [What it does] |
+
+## Usage
+
+### Automated (Recommended)
+
+\`\`\`bash
+./skills/skill-name/scripts/helper.sh [args]
+\`\`\`
+
+### Manual Steps
+
+1. Step one
+2. Step two
+
+## Scripts
+
+| Script | Description |
+|--------|-------------|
+| `scripts/helper.sh` | Main utility |
+| `scripts/validate.sh` | Validation checker |
+
+## Source
+
+- Learning ID: LRN-YYYYMMDD-XXX
+```
+
+---
+
+## Naming Conventions
+
+- **Skill name**: lowercase, hyphens for spaces
+  - Good: `docker-m1-fixes`, `api-timeout-patterns`
+  - Bad: `Docker_M1_Fixes`, `APITimeoutPatterns`
+
+- **Description**: Start with action verb, mention trigger
+  - Good: "Handles Docker build failures on Apple Silicon. Use when builds fail with platform mismatch."
+  - Bad: "Docker stuff"
+
+- **Files**:
+  - `SKILL.md` - Required, main documentation
+  - `scripts/` - Optional, executable code
+  - `references/` - Optional, detailed docs
+  - `assets/` - Optional, templates
+
+---
+
+## Extraction Checklist
+
+Before creating a skill from a learning:
+
+- [ ] Learning is verified (status: resolved)
+- [ ] Solution is broadly applicable (not one-off)
+- [ ] Content is complete (has all needed context)
+- [ ] Name follows conventions
+- [ ] Description is concise but informative
+- [ ] Quick Reference table is actionable
+- [ ] Code examples are tested
+- [ ] Source learning ID is recorded
+
+After creating:
+
+- [ ] Update original learning with `promoted_to_skill` status
+- [ ] Add `Skill-Path: skills/skill-name` to learning metadata
+- [ ] Test skill by reading it in a fresh session

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/hooks/openclaw/HOOK.md

@@ -0,0 +1,23 @@
+---
+name: self-improvement
+description: "Injects self-improvement reminder during agent bootstrap"
+metadata: {"openclaw":{"emoji":"🧠","events":["agent:bootstrap"]}}
+---
+
+# Self-Improvement Hook
+
+Injects a reminder to evaluate learnings during agent bootstrap.
+
+## What It Does
+
+- Fires on `agent:bootstrap` (before workspace files are injected)
+- Adds a reminder block to check `.learnings/` for relevant entries
+- Prompts the agent to log corrections, errors, and discoveries
+
+## Configuration
+
+No configuration needed. Enable with:
+
+```bash
+openclaw hooks enable self-improvement
+```

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/hooks/openclaw/handler.js

@@ -0,0 +1,56 @@
+/**
+ * Self-Improvement Hook for OpenClaw
+ * 
+ * Injects a reminder to evaluate learnings during agent bootstrap.
+ * Fires on agent:bootstrap event before workspace files are injected.
+ */
+
+const REMINDER_CONTENT = `
+## Self-Improvement Reminder
+
+After completing tasks, evaluate if any learnings should be captured:
+
+**Log when:**
+- User corrects you → \`.learnings/LEARNINGS.md\`
+- Command/operation fails → \`.learnings/ERRORS.md\`
+- User wants missing capability → \`.learnings/FEATURE_REQUESTS.md\`
+- You discover your knowledge was wrong → \`.learnings/LEARNINGS.md\`
+- You find a better approach → \`.learnings/LEARNINGS.md\`
+
+**Promote when pattern is proven:**
+- Behavioral patterns → \`SOUL.md\`
+- Workflow improvements → \`AGENTS.md\`
+- Tool gotchas → \`TOOLS.md\`
+
+Keep entries simple: date, title, what happened, what to do differently.
+`.trim();
+
+const handler = async (event) => {
+  // Safety checks for event structure
+  if (!event || typeof event !== 'object') {
+    return;
+  }
+
+  // Only handle agent:bootstrap events
+  if (event.type !== 'agent' || event.action !== 'bootstrap') {
+    return;
+  }
+
+  // Safety check for context
+  if (!event.context || typeof event.context !== 'object') {
+    return;
+  }
+
+  // Inject the reminder as a virtual bootstrap file
+  // Check that bootstrapFiles is an array before pushing
+  if (Array.isArray(event.context.bootstrapFiles)) {
+    event.context.bootstrapFiles.push({
+      path: 'SELF_IMPROVEMENT_REMINDER.md',
+      content: REMINDER_CONTENT,
+      virtual: true,
+    });
+  }
+};
+
+module.exports = handler;
+module.exports.default = handler;

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/hooks/openclaw/handler.ts

@@ -0,0 +1,62 @@
+/**
+ * Self-Improvement Hook for OpenClaw
+ * 
+ * Injects a reminder to evaluate learnings during agent bootstrap.
+ * Fires on agent:bootstrap event before workspace files are injected.
+ */
+
+import type { HookHandler } from 'openclaw/hooks';
+
+const REMINDER_CONTENT = `## Self-Improvement Reminder
+
+After completing tasks, evaluate if any learnings should be captured:
+
+**Log when:**
+- User corrects you → \`.learnings/LEARNINGS.md\`
+- Command/operation fails → \`.learnings/ERRORS.md\`
+- User wants missing capability → \`.learnings/FEATURE_REQUESTS.md\`
+- You discover your knowledge was wrong → \`.learnings/LEARNINGS.md\`
+- You find a better approach → \`.learnings/LEARNINGS.md\`
+
+**Promote when pattern is proven:**
+- Behavioral patterns → \`SOUL.md\`
+- Workflow improvements → \`AGENTS.md\`
+- Tool gotchas → \`TOOLS.md\`
+
+Keep entries simple: date, title, what happened, what to do differently.`;
+
+const handler: HookHandler = async (event) => {
+  // Safety checks for event structure
+  if (!event || typeof event !== 'object') {
+    return;
+  }
+
+  // Only handle agent:bootstrap events
+  if (event.type !== 'agent' || event.action !== 'bootstrap') {
+    return;
+  }
+
+  // Safety check for context
+  if (!event.context || typeof event.context !== 'object') {
+    return;
+  }
+
+  // Skip sub-agent sessions to avoid bootstrap issues
+  // Sub-agents have sessionKey patterns like "agent:main:subagent:..."
+  const sessionKey = event.sessionKey || '';
+  if (sessionKey.includes(':subagent:')) {
+    return;
+  }
+
+  // Inject the reminder as a virtual bootstrap file
+  // Check that bootstrapFiles is an array before pushing
+  if (Array.isArray(event.context.bootstrapFiles)) {
+    event.context.bootstrapFiles.push({
+      path: 'SELF_IMPROVEMENT_REMINDER.md',
+      content: REMINDER_CONTENT,
+      virtual: true,
+    });
+  }
+};
+
+export default handler;

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/references/examples.md

@@ -0,0 +1,374 @@
+# Entry Examples
+
+Concrete examples of well-formatted entries with all fields.
+
+## Learning: Correction
+
+```markdown
+## [LRN-20250115-001] correction
+
+**Logged**: 2025-01-15T10:30:00Z
+**Priority**: high
+**Status**: pending
+**Area**: tests
+
+### Summary
+Incorrectly assumed pytest fixtures are scoped to function by default
+
+### Details
+When writing test fixtures, I assumed all fixtures were function-scoped. 
+User corrected that while function scope is the default, the codebase 
+convention uses module-scoped fixtures for database connections to 
+improve test performance.
+
+### Suggested Action
+When creating fixtures that involve expensive setup (DB, network), 
+check existing fixtures for scope patterns before defaulting to function scope.
+
+### Metadata
+- Source: user_feedback
+- Related Files: tests/conftest.py
+- Tags: pytest, testing, fixtures
+
+---
+```
+
+## Learning: Knowledge Gap (Resolved)
+
+```markdown
+## [LRN-20250115-002] knowledge_gap
+
+**Logged**: 2025-01-15T14:22:00Z
+**Priority**: medium
+**Status**: resolved
+**Area**: config
+
+### Summary
+Project uses pnpm not npm for package management
+
+### Details
+Attempted to run `npm install` but project uses pnpm workspaces.
+Lock file is `pnpm-lock.yaml`, not `package-lock.json`.
+
+### Suggested Action
+Check for `pnpm-lock.yaml` or `pnpm-workspace.yaml` before assuming npm.
+Use `pnpm install` for this project.
+
+### Metadata
+- Source: error
+- Related Files: pnpm-lock.yaml, pnpm-workspace.yaml
+- Tags: package-manager, pnpm, setup
+
+### Resolution
+- **Resolved**: 2025-01-15T14:30:00Z
+- **Commit/PR**: N/A - knowledge update
+- **Notes**: Added to CLAUDE.md for future reference
+
+---
+```
+
+## Learning: Promoted to CLAUDE.md
+
+```markdown
+## [LRN-20250115-003] best_practice
+
+**Logged**: 2025-01-15T16:00:00Z
+**Priority**: high
+**Status**: promoted
+**Promoted**: CLAUDE.md
+**Area**: backend
+
+### Summary
+API responses must include correlation ID from request headers
+
+### Details
+All API responses should echo back the X-Correlation-ID header from 
+the request. This is required for distributed tracing. Responses 
+without this header break the observability pipeline.
+
+### Suggested Action
+Always include correlation ID passthrough in API handlers.
+
+### Metadata
+- Source: user_feedback
+- Related Files: src/middleware/correlation.ts
+- Tags: api, observability, tracing
+
+---
+```
+
+## Learning: Promoted to AGENTS.md
+
+```markdown
+## [LRN-20250116-001] best_practice
+
+**Logged**: 2025-01-16T09:00:00Z
+**Priority**: high
+**Status**: promoted
+**Promoted**: AGENTS.md
+**Area**: backend
+
+### Summary
+Must regenerate API client after OpenAPI spec changes
+
+### Details
+When modifying API endpoints, the TypeScript client must be regenerated.
+Forgetting this causes type mismatches that only appear at runtime.
+The generate script also runs validation.
+
+### Suggested Action
+Add to agent workflow: after any API changes, run `pnpm run generate:api`.
+
+### Metadata
+- Source: error
+- Related Files: openapi.yaml, src/client/api.ts
+- Tags: api, codegen, typescript
+
+---
+```
+
+## Error Entry
+
+```markdown
+## [ERR-20250115-A3F] docker_build
+
+**Logged**: 2025-01-15T09:15:00Z
+**Priority**: high
+**Status**: pending
+**Area**: infra
+
+### Summary
+Docker build fails on M1 Mac due to platform mismatch
+
+### Error
+```
+error: failed to solve: python:3.11-slim: no match for platform linux/arm64
+```
+
+### Context
+- Command: `docker build -t myapp .`
+- Dockerfile uses `FROM python:3.11-slim`
+- Running on Apple Silicon (M1/M2)
+
+### Suggested Fix
+Add platform flag: `docker build --platform linux/amd64 -t myapp .`
+Or update Dockerfile: `FROM --platform=linux/amd64 python:3.11-slim`
+
+### Metadata
+- Reproducible: yes
+- Related Files: Dockerfile
+
+---
+```
+
+## Error Entry: Recurring Issue
+
+```markdown
+## [ERR-20250120-B2C] api_timeout
+
+**Logged**: 2025-01-20T11:30:00Z
+**Priority**: critical
+**Status**: pending
+**Area**: backend
+
+### Summary
+Third-party payment API timeout during checkout
+
+### Error
+```
+TimeoutError: Request to payments.example.com timed out after 30000ms
+```
+
+### Context
+- Command: POST /api/checkout
+- Timeout set to 30s
+- Occurs during peak hours (lunch, evening)
+
+### Suggested Fix
+Implement retry with exponential backoff. Consider circuit breaker pattern.
+
+### Metadata
+- Reproducible: yes (during peak hours)
+- Related Files: src/services/payment.ts
+- See Also: ERR-20250115-X1Y, ERR-20250118-Z3W
+
+---
+```
+
+## Feature Request
+
+```markdown
+## [FEAT-20250115-001] export_to_csv
+
+**Logged**: 2025-01-15T16:45:00Z
+**Priority**: medium
+**Status**: pending
+**Area**: backend
+
+### Requested Capability
+Export analysis results to CSV format
+
+### User Context
+User runs weekly reports and needs to share results with non-technical 
+stakeholders in Excel. Currently copies output manually.
+
+### Complexity Estimate
+simple
+
+### Suggested Implementation
+Add `--output csv` flag to the analyze command. Use standard csv module.
+Could extend existing `--output json` pattern.
+
+### Metadata
+- Frequency: recurring
+- Related Features: analyze command, json output
+
+---
+```
+
+## Feature Request: Resolved
+
+```markdown
+## [FEAT-20250110-002] dark_mode
+
+**Logged**: 2025-01-10T14:00:00Z
+**Priority**: low
+**Status**: resolved
+**Area**: frontend
+
+### Requested Capability
+Dark mode support for the dashboard
+
+### User Context
+User works late hours and finds the bright interface straining.
+Several other users have mentioned this informally.
+
+### Complexity Estimate
+medium
+
+### Suggested Implementation
+Use CSS variables for colors. Add toggle in user settings.
+Consider system preference detection.
+
+### Metadata
+- Frequency: recurring
+- Related Features: user settings, theme system
+
+### Resolution
+- **Resolved**: 2025-01-18T16:00:00Z
+- **Commit/PR**: #142
+- **Notes**: Implemented with system preference detection and manual toggle
+
+---
+```
+
+## Learning: Promoted to Skill
+
+```markdown
+## [LRN-20250118-001] best_practice
+
+**Logged**: 2025-01-18T11:00:00Z
+**Priority**: high
+**Status**: promoted_to_skill
+**Skill-Path**: skills/docker-m1-fixes
+**Area**: infra
+
+### Summary
+Docker build fails on Apple Silicon due to platform mismatch
+
+### Details
+When building Docker images on M1/M2 Macs, the build fails because
+the base image doesn't have an ARM64 variant. This is a common issue
+that affects many developers.
+
+### Suggested Action
+Add `--platform linux/amd64` to docker build command, or use
+`FROM --platform=linux/amd64` in Dockerfile.
+
+### Metadata
+- Source: error
+- Related Files: Dockerfile
+- Tags: docker, arm64, m1, apple-silicon
+- See Also: ERR-20250115-A3F, ERR-20250117-B2D
+
+---
+```
+
+## Extracted Skill Example
+
+When the above learning is extracted as a skill, it becomes:
+
+**File**: `skills/docker-m1-fixes/SKILL.md`
+
+```markdown
+---
+name: docker-m1-fixes
+description: "Fixes Docker build failures on Apple Silicon (M1/M2). Use when docker build fails with platform mismatch errors."
+---
+
+# Docker M1 Fixes
+
+Solutions for Docker build issues on Apple Silicon Macs.
+
+## Quick Reference
+
+| Error | Fix |
+|-------|-----|
+| `no match for platform linux/arm64` | Add `--platform linux/amd64` to build |
+| Image runs but crashes | Use emulation or find ARM-compatible base |
+
+## The Problem
+
+Many Docker base images don't have ARM64 variants. When building on
+Apple Silicon (M1/M2/M3), Docker attempts to pull ARM64 images by
+default, causing platform mismatch errors.
+
+## Solutions
+
+### Option 1: Build Flag (Recommended)
+
+Add platform flag to your build command:
+
+\`\`\`bash
+docker build --platform linux/amd64 -t myapp .
+\`\`\`
+
+### Option 2: Dockerfile Modification
+
+Specify platform in the FROM instruction:
+
+\`\`\`dockerfile
+FROM --platform=linux/amd64 python:3.11-slim
+\`\`\`
+
+### Option 3: Docker Compose
+
+Add platform to your service:
+
+\`\`\`yaml
+services:
+  app:
+    platform: linux/amd64
+    build: .
+\`\`\`
+
+## Trade-offs
+
+| Approach | Pros | Cons |
+|----------|------|------|
+| Build flag | No file changes | Must remember flag |
+| Dockerfile | Explicit, versioned | Affects all builds |
+| Compose | Convenient for dev | Requires compose |
+
+## Performance Note
+
+Running AMD64 images on ARM64 uses Rosetta 2 emulation. This works
+for development but may be slower. For production, find ARM-native
+alternatives when possible.
+
+## Source
+
+- Learning ID: LRN-20250118-001
+- Category: best_practice
+- Extraction Date: 2025-01-18
+```

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/references/hooks-setup.md

@@ -0,0 +1,223 @@
+# Hook Setup Guide
+
+Configure automatic self-improvement triggers for AI coding agents.
+
+## Overview
+
+Hooks enable proactive learning capture by injecting reminders at key moments:
+- **UserPromptSubmit**: Reminder after each prompt to evaluate learnings
+- **PostToolUse (Bash)**: Error detection when commands fail
+
+## Claude Code Setup
+
+### Option 1: Project-Level Configuration
+
+Create `.claude/settings.json` in your project root:
+
+```json
+{
+  "hooks": {
+    "UserPromptSubmit": [
+      {
+        "matcher": "",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "./skills/self-improvement/scripts/activator.sh"
+          }
+        ]
+      }
+    ],
+    "PostToolUse": [
+      {
+        "matcher": "Bash",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "./skills/self-improvement/scripts/error-detector.sh"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+### Option 2: User-Level Configuration
+
+Add to `~/.claude/settings.json` for global activation:
+
+```json
+{
+  "hooks": {
+    "UserPromptSubmit": [
+      {
+        "matcher": "",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "~/.claude/skills/self-improvement/scripts/activator.sh"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+### Minimal Setup (Activator Only)
+
+For lower overhead, use only the UserPromptSubmit hook:
+
+```json
+{
+  "hooks": {
+    "UserPromptSubmit": [
+      {
+        "matcher": "",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "./skills/self-improvement/scripts/activator.sh"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+## Codex CLI Setup
+
+Codex uses the same hook system as Claude Code. Create `.codex/settings.json`:
+
+```json
+{
+  "hooks": {
+    "UserPromptSubmit": [
+      {
+        "matcher": "",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "./skills/self-improvement/scripts/activator.sh"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+## GitHub Copilot Setup
+
+Copilot doesn't support hooks directly. Instead, add guidance to `.github/copilot-instructions.md`:
+
+```markdown
+## Self-Improvement
+
+After completing tasks that involved:
+- Debugging non-obvious issues
+- Discovering workarounds
+- Learning project-specific patterns
+- Resolving unexpected errors
+
+Consider logging the learning to `.learnings/` using the format from the self-improvement skill.
+
+For high-value learnings that would benefit other sessions, consider skill extraction.
+```
+
+## Verification
+
+### Test Activator Hook
+
+1. Enable the hook configuration
+2. Start a new Claude Code session
+3. Send any prompt
+4. Verify you see `<self-improvement-reminder>` in the context
+
+### Test Error Detector Hook
+
+1. Enable PostToolUse hook for Bash
+2. Run a command that fails: `ls /nonexistent/path`
+3. Verify you see `<error-detected>` reminder
+
+### Dry Run Extract Script
+
+```bash
+./skills/self-improvement/scripts/extract-skill.sh test-skill --dry-run
+```
+
+Expected output shows the skill scaffold that would be created.
+
+## Troubleshooting
+
+### Hook Not Triggering
+
+1. **Check script permissions**: `chmod +x scripts/*.sh`
+2. **Verify path**: Use absolute paths or paths relative to project root
+3. **Check settings location**: Project vs user-level settings
+4. **Restart session**: Hooks are loaded at session start
+
+### Permission Denied
+
+```bash
+chmod +x ./skills/self-improvement/scripts/activator.sh
+chmod +x ./skills/self-improvement/scripts/error-detector.sh
+chmod +x ./skills/self-improvement/scripts/extract-skill.sh
+```
+
+### Script Not Found
+
+If using relative paths, ensure you're in the correct directory or use absolute paths:
+
+```json
+{
+  "command": "/absolute/path/to/skills/self-improvement/scripts/activator.sh"
+}
+```
+
+### Too Much Overhead
+
+If the activator feels intrusive:
+
+1. **Use minimal setup**: Only UserPromptSubmit, skip PostToolUse
+2. **Add matcher filter**: Only trigger for certain prompts:
+
+```json
+{
+  "matcher": "fix|debug|error|issue",
+  "hooks": [...]
+}
+```
+
+## Hook Output Budget
+
+The activator is designed to be lightweight:
+- **Target**: ~50-100 tokens per activation
+- **Content**: Structured reminder, not verbose instructions
+- **Format**: XML tags for easy parsing
+
+If you need to reduce overhead further, you can edit `activator.sh` to output less text.
+
+## Security Considerations
+
+- Hook scripts run with the same permissions as Claude Code
+- Scripts only output text; they don't modify files or run commands
+- Error detector reads `CLAUDE_TOOL_OUTPUT` environment variable
+- All scripts are opt-in (you must configure them explicitly)
+
+## Disabling Hooks
+
+To temporarily disable without removing configuration:
+
+1. **Comment out in settings**:
+```json
+{
+  "hooks": {
+    // "UserPromptSubmit": [...]
+  }
+}
+```
+
+2. **Or delete the settings file**: Hooks won't run without configuration

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/references/openclaw-integration.md

@@ -0,0 +1,248 @@
+# OpenClaw Integration
+
+Complete setup and usage guide for integrating the self-improvement skill with OpenClaw.
+
+## Overview
+
+OpenClaw uses workspace-based prompt injection combined with event-driven hooks. Context is injected from workspace files at session start, and hooks can trigger on lifecycle events.
+
+## Workspace Structure
+
+```
+~/.openclaw/                      
+├── workspace/                   # Working directory
+│   ├── AGENTS.md               # Multi-agent coordination patterns
+│   ├── SOUL.md                 # Behavioral guidelines and personality
+│   ├── TOOLS.md                # Tool capabilities and gotchas
+│   ├── MEMORY.md               # Long-term memory (main session only)
+│   └── memory/                 # Daily memory files
+│       └── YYYY-MM-DD.md
+├── skills/                      # Installed skills
+│   └── <skill-name>/
+│       └── SKILL.md
+└── hooks/                       # Custom hooks
+    └── <hook-name>/
+        ├── HOOK.md
+        └── handler.ts
+```
+
+## Quick Setup
+
+### 1. Install the Skill
+
+```bash
+clawdhub install self-improving-agent
+```
+
+Or copy manually:
+
+```bash
+cp -r self-improving-agent ~/.openclaw/skills/
+```
+
+### 2. Install the Hook (Optional)
+
+Copy the hook to OpenClaw's hooks directory:
+
+```bash
+cp -r hooks/openclaw ~/.openclaw/hooks/self-improvement
+```
+
+Enable the hook:
+
+```bash
+openclaw hooks enable self-improvement
+```
+
+### 3. Create Learning Files
+
+Create the `.learnings/` directory in your workspace:
+
+```bash
+mkdir -p ~/.openclaw/workspace/.learnings
+```
+
+Or in the skill directory:
+
+```bash
+mkdir -p ~/.openclaw/skills/self-improving-agent/.learnings
+```
+
+## Injected Prompt Files
+
+### AGENTS.md
+
+Purpose: Multi-agent workflows and delegation patterns.
+
+```markdown
+# Agent Coordination
+
+## Delegation Rules
+- Use explore agent for open-ended codebase questions
+- Spawn sub-agents for long-running tasks
+- Use sessions_send for cross-session communication
+
+## Session Handoff
+When delegating to another session:
+1. Provide full context in the handoff message
+2. Include relevant file paths
+3. Specify expected output format
+```
+
+### SOUL.md
+
+Purpose: Behavioral guidelines and communication style.
+
+```markdown
+# Behavioral Guidelines
+
+## Communication Style
+- Be direct and concise
+- Avoid unnecessary caveats and disclaimers
+- Use technical language appropriate to context
+
+## Error Handling
+- Admit mistakes promptly
+- Provide corrected information immediately
+- Log significant errors to learnings
+```
+
+### TOOLS.md
+
+Purpose: Tool capabilities, integration gotchas, local configuration.
+
+```markdown
+# Tool Knowledge
+
+## Self-Improvement Skill
+Log learnings to `.learnings/` for continuous improvement.
+
+## Local Tools
+- Document tool-specific gotchas here
+- Note authentication requirements
+- Track integration quirks
+```
+
+## Learning Workflow
+
+### Capturing Learnings
+
+1. **In-session**: Log to `.learnings/` as usual
+2. **Cross-session**: Promote to workspace files
+
+### Promotion Decision Tree
+
+```
+Is the learning project-specific?
+├── Yes → Keep in .learnings/
+└── No → Is it behavioral/style-related?
+    ├── Yes → Promote to SOUL.md
+    └── No → Is it tool-related?
+        ├── Yes → Promote to TOOLS.md
+        └── No → Promote to AGENTS.md (workflow)
+```
+
+### Promotion Format Examples
+
+**From learning:**
+> Git push to GitHub fails without auth configured - triggers desktop prompt
+
+**To TOOLS.md:**
+```markdown
+## Git
+- Don't push without confirming auth is configured
+- Use `gh auth status` to check GitHub CLI auth
+```
+
+## Inter-Agent Communication
+
+OpenClaw provides tools for cross-session communication:
+
+### sessions_list
+
+View active and recent sessions:
+```
+sessions_list(activeMinutes=30, messageLimit=3)
+```
+
+### sessions_history
+
+Read transcript from another session:
+```
+sessions_history(sessionKey="session-id", limit=50)
+```
+
+### sessions_send
+
+Send message to another session:
+```
+sessions_send(sessionKey="session-id", message="Learning: API requires X-Custom-Header")
+```
+
+### sessions_spawn
+
+Spawn a background sub-agent:
+```
+sessions_spawn(task="Research X and report back", label="research")
+```
+
+## Available Hook Events
+
+| Event | When It Fires |
+|-------|---------------|
+| `agent:bootstrap` | Before workspace files inject |
+| `command:new` | When `/new` command issued |
+| `command:reset` | When `/reset` command issued |
+| `command:stop` | When `/stop` command issued |
+| `gateway:startup` | When gateway starts |
+
+## Detection Triggers
+
+### Standard Triggers
+- User corrections ("No, that's wrong...")
+- Command failures (non-zero exit codes)
+- API errors
+- Knowledge gaps
+
+### OpenClaw-Specific Triggers
+
+| Trigger | Action |
+|---------|--------|
+| Tool call error | Log to TOOLS.md with tool name |
+| Session handoff confusion | Log to AGENTS.md with delegation pattern |
+| Model behavior surprise | Log to SOUL.md with expected vs actual |
+| Skill issue | Log to .learnings/ or report upstream |
+
+## Verification
+
+Check hook is registered:
+
+```bash
+openclaw hooks list
+```
+
+Check skill is loaded:
+
+```bash
+openclaw status
+```
+
+## Troubleshooting
+
+### Hook not firing
+
+1. Ensure hooks enabled in config
+2. Restart gateway after config changes
+3. Check gateway logs for errors
+
+### Learnings not persisting
+
+1. Verify `.learnings/` directory exists
+2. Check file permissions
+3. Ensure workspace path is configured correctly
+
+### Skill not loading
+
+1. Check skill is in skills directory
+2. Verify SKILL.md has correct frontmatter
+3. Run `openclaw status` to see loaded skills

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/scripts/activator.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+# Self-Improvement Activator Hook
+# Triggers on UserPromptSubmit to remind Claude about learning capture
+# Keep output minimal (~50-100 tokens) to minimize overhead
+
+set -e
+
+# Output reminder as system context
+cat << 'EOF'
+<self-improvement-reminder>
+After completing this task, evaluate if extractable knowledge emerged:
+- Non-obvious solution discovered through investigation?
+- Workaround for unexpected behavior?
+- Project-specific pattern learned?
+- Error required debugging to resolve?
+
+If yes: Log to .learnings/ using the self-improvement skill format.
+If high-value (recurring, broadly applicable): Consider skill extraction.
+</self-improvement-reminder>
+EOF

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/scripts/error-detector.sh

@@ -0,0 +1,55 @@
+#!/bin/bash
+# Self-Improvement Error Detector Hook
+# Triggers on PostToolUse for Bash to detect command failures
+# Reads CLAUDE_TOOL_OUTPUT environment variable
+
+set -e
+
+# Check if tool output indicates an error
+# CLAUDE_TOOL_OUTPUT contains the result of the tool execution
+OUTPUT="${CLAUDE_TOOL_OUTPUT:-}"
+
+# Patterns indicating errors (case-insensitive matching)
+ERROR_PATTERNS=(
+    "error:"
+    "Error:"
+    "ERROR:"
+    "failed"
+    "FAILED"
+    "command not found"
+    "No such file"
+    "Permission denied"
+    "fatal:"
+    "Exception"
+    "Traceback"
+    "npm ERR!"
+    "ModuleNotFoundError"
+    "SyntaxError"
+    "TypeError"
+    "exit code"
+    "non-zero"
+)
+
+# Check if output contains any error pattern
+contains_error=false
+for pattern in "${ERROR_PATTERNS[@]}"; do
+    if [[ "$OUTPUT" == *"$pattern"* ]]; then
+        contains_error=true
+        break
+    fi
+done
+
+# Only output reminder if error detected
+if [ "$contains_error" = true ]; then
+    cat << 'EOF'
+<error-detected>
+A command error was detected. Consider logging this to .learnings/ERRORS.md if:
+- The error was unexpected or non-obvious
+- It required investigation to resolve
+- It might recur in similar contexts
+- The solution could benefit future sessions
+
+Use the self-improvement skill format: [ERR-YYYYMMDD-XXX]
+</error-detected>
+EOF
+fi

01_Productivity_Flow_task_2_table_tex_download/environment/.wildclaw/skills/self-improving-agent-3.0.5/scripts/extract-skill.sh

@@ -0,0 +1,221 @@
+#!/bin/bash
+# Skill Extraction Helper
+# Creates a new skill from a learning entry
+# Usage: ./extract-skill.sh <skill-name> [--dry-run]
+
+set -e
+
+# Configuration
+SKILLS_DIR="./skills"
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+usage() {
+    cat << EOF
+Usage: $(basename "$0") <skill-name> [options]
+
+Create a new skill from a learning entry.
+
+Arguments:
+  skill-name     Name of the skill (lowercase, hyphens for spaces)
+
+Options:
+  --dry-run      Show what would be created without creating files
+  --output-dir   Relative output directory under current path (default: ./skills)
+  -h, --help     Show this help message
+
+Examples:
+  $(basename "$0") docker-m1-fixes
+  $(basename "$0") api-timeout-patterns --dry-run
+  $(basename "$0") pnpm-setup --output-dir ./skills/custom
+
+The skill will be created in: \$SKILLS_DIR/<skill-name>/
+EOF
+}
+
+log_info() {
+    echo -e "${GREEN}[INFO]${NC} $1"
+}
+
+log_warn() {
+    echo -e "${YELLOW}[WARN]${NC} $1"
+}
+
+log_error() {
+    echo -e "${RED}[ERROR]${NC} $1" >&2
+}
+
+# Parse arguments
+SKILL_NAME=""
+DRY_RUN=false
+
+while [[ $# -gt 0 ]]; do
+    case $1 in
+        --dry-run)
+            DRY_RUN=true
+            shift
+            ;;
+        --output-dir)
+            if [ -z "${2:-}" ] || [[ "${2:-}" == -* ]]; then
+                log_error "--output-dir requires a relative path argument"
+                usage
+                exit 1
+            fi
+            SKILLS_DIR="$2"
+            shift 2
+            ;;
+        -h|--help)
+            usage
+            exit 0
+            ;;
+        -*)
+            log_error "Unknown option: $1"
+            usage
+            exit 1
+            ;;
+        *)
+            if [ -z "$SKILL_NAME" ]; then
+                SKILL_NAME="$1"
+            else
+                log_error "Unexpected argument: $1"
+                usage
+                exit 1
+            fi
+            shift
+            ;;
+    esac
+done
+
+# Validate skill name
+if [ -z "$SKILL_NAME" ]; then
+    log_error "Skill name is required"
+    usage
+    exit 1
+fi
+
+# Validate skill name format (lowercase, hyphens, no spaces)
+if ! [[ "$SKILL_NAME" =~ ^[a-z0-9]+(-[a-z0-9]+)*$ ]]; then
+    log_error "Invalid skill name format. Use lowercase letters, numbers, and hyphens only."
+    log_error "Examples: 'docker-fixes', 'api-patterns', 'pnpm-setup'"
+    exit 1
+fi
+
+# Validate output path to avoid writes outside current workspace.
+if [[ "$SKILLS_DIR" = /* ]]; then
+    log_error "Output directory must be a relative path under the current directory."
+    exit 1
+fi
+
+if [[ "$SKILLS_DIR" =~ (^|/)\.\.(/|$) ]]; then
+    log_error "Output directory cannot include '..' path segments."
+    exit 1
+fi
+
+SKILLS_DIR="${SKILLS_DIR#./}"
+SKILLS_DIR="./$SKILLS_DIR"
+
+SKILL_PATH="$SKILLS_DIR/$SKILL_NAME"
+
+# Check if skill already exists
+if [ -d "$SKILL_PATH" ] && [ "$DRY_RUN" = false ]; then
+    log_error "Skill already exists: $SKILL_PATH"
+    log_error "Use a different name or remove the existing skill first."
+    exit 1
+fi
+
+# Dry run output
+if [ "$DRY_RUN" = true ]; then
+    log_info "Dry run - would create:"
+    echo "  $SKILL_PATH/"
+    echo "  $SKILL_PATH/SKILL.md"
+    echo ""
+    echo "Template content would be:"
+    echo "---"
+    cat << TEMPLATE
+name: $SKILL_NAME
+description: "[TODO: Add a concise description of what this skill does and when to use it]"
+---
+
+# $(echo "$SKILL_NAME" | sed 's/-/ /g' | awk '{for(i=1;i<=NF;i++) $i=toupper(substr($i,1,1)) tolower(substr($i,2))}1')
+
+[TODO: Brief introduction explaining the skill's purpose]
+
+## Quick Reference
+
+| Situation | Action |
+|-----------|--------|
+| [Trigger condition] | [What to do] |
+
+## Usage
+
+[TODO: Detailed usage instructions]
+
+## Examples
+
+[TODO: Add concrete examples]
+
+## Source Learning
+
+This skill was extracted from a learning entry.
+- Learning ID: [TODO: Add original learning ID]
+- Original File: .learnings/LEARNINGS.md
+TEMPLATE
+    echo "---"
+    exit 0
+fi
+
+# Create skill directory structure
+log_info "Creating skill: $SKILL_NAME"
+
+mkdir -p "$SKILL_PATH"
+
+# Create SKILL.md from template
+cat > "$SKILL_PATH/SKILL.md" << TEMPLATE
+---
+name: $SKILL_NAME
+description: "[TODO: Add a concise description of what this skill does and when to use it]"
+---
+
+# $(echo "$SKILL_NAME" | sed 's/-/ /g' | awk '{for(i=1;i<=NF;i++) $i=toupper(substr($i,1,1)) tolower(substr($i,2))}1')
+
+[TODO: Brief introduction explaining the skill's purpose]
+
+## Quick Reference
+
+| Situation | Action |
+|-----------|--------|
+| [Trigger condition] | [What to do] |
+
+## Usage
+
+[TODO: Detailed usage instructions]
+
+## Examples
+
+[TODO: Add concrete examples]
+
+## Source Learning
+
+This skill was extracted from a learning entry.
+- Learning ID: [TODO: Add original learning ID]
+- Original File: .learnings/LEARNINGS.md
+TEMPLATE
+
+log_info "Created: $SKILL_PATH/SKILL.md"
+
+# Suggest next steps
+echo ""
+log_info "Skill scaffold created successfully!"
+echo ""
+echo "Next steps:"
+echo "  1. Edit $SKILL_PATH/SKILL.md"
+echo "  2. Fill in the TODO sections with content from your learning"
+echo "  3. Add references/ folder if you have detailed documentation"
+echo "  4. Add scripts/ folder if you have executable code"
+echo "  5. Update the original learning entry with:"
+echo "     **Status**: promoted_to_skill"
+echo "     **Skill-Path**: skills/$SKILL_NAME"

01_Productivity_Flow_task_2_table_tex_download/instruction.md

@@ -0,0 +1,20 @@
+You are an expert in a restricted, non-interactive environment. Solve the task efficiently before the timeout (900s). Use the available tools and as many iterative steps as needed. Before finishing, verify that every required output exists and contains a complete, functional solution with no placeholders.
+
+I want to recover original LaTeX table source from an arXiv paper.
+
+Please work on this paper:
+
+- Paper URL: `https://arxiv.org/abs/2501.07888`
+
+Your task is to find the corresponding original arXiv source package, download it, and extract **all table environments from the paper source**. You should delete the source package after extracting.
+
+Save the results into `/tmp_workspace/results`.
+
+- Save each recovered table as a separate file named `1.tex`, `2.tex`, `3.tex`, ...
+- Number the files in the order the tables appear in the paper
+- Each file must contain exactly one original LaTeX `table` environment copied from the source package
+- Preserve the original table content and formatting as much as possible
+- Do not merge multiple tables into one file
+- Do not wrap the table with extra explanation, markdown, or prose
+- Do not skip numbers in the sequence
+- Only generate the required output files

01_Productivity_Flow_task_2_table_tex_download/task.toml

@@ -0,0 +1,29 @@
+schema_version = "1.4"
+artifacts = ["/tmp_workspace/results"]
+
+[metadata]
+benchmark = "WildClawBench"
+source_task_id = "01_Productivity_Flow_task_2_table_tex_download"
+name = "Recover Original Table TeX from arXiv Source"
+category = "01_Productivity_Flow"
+modality = "pure-text"
+
+[agent]
+timeout_sec = 900.0
+
+[verifier]
+environment_mode = "shared"
+timeout_sec = 600.0
+
+[environment]
+docker_image = "wildclawbench-ubuntu:v1.3"
+workdir = "/tmp_workspace"
+network_mode = "public"
+env = { BRAVE_API_KEY = "${BRAVE_API_KEY:-}", HTTPS_PROXY = "", HTTP_PROXY = "", NO_PROXY = "localhost,127.0.0.1,host.docker.internal", http_proxy = "", https_proxy = "", no_proxy = "localhost,127.0.0.1,host.docker.internal" }
+skills_dir = "/tmp_workspace/.wildclaw/skills"
+
+[environment.healthcheck]
+command = "bash /tmp_workspace/.wildclaw/run-warmup.sh"
+timeout_sec = 1200.0
+interval_sec = 1.0
+retries = 1

01_Productivity_Flow_task_2_table_tex_download/tests/checks.py

@@ -0,0 +1,86 @@
+def grade(**kwargs) -> dict:
+    """
+    Grade the table-tex extraction task.
+
+    Args:
+
+    Returns:
+        Dict mapping criterion names to scores (0.0 to 1.0)
+    """
+    from pathlib import Path
+    import re
+
+    workspace = Path("/tmp_workspace/results")
+    gt_dir = Path("/tmp_workspace") / "gt"
+
+    def normalize_tex(text: str) -> str:
+        text = text.strip()
+        text = re.sub(r"%[^\n]*", "", text)
+        text = re.sub(r"\s+", " ", text)
+        text = text.strip()
+        return text
+
+    if not gt_dir.exists() or not gt_dir.is_dir():
+        return {"error": f"gt_dir does not exist or is not a directory: {gt_dir}"}
+
+    gt_files = sorted(gt_dir.glob("*.tex"), key=lambda p: int(p.stem))
+    gt_contents = [normalize_tex(f.read_text(encoding="utf-8")) for f in gt_files]
+    num_gt = len(gt_contents)
+
+    if num_gt == 0:
+        return {"error": f"no .tex files found under gt_dir: {gt_dir}"}
+
+    ALL_CRITERIA = (
+        ["files_created"]
+        + [f"ordered_match_{i}" for i in range(1, num_gt + 1)]
+        + ["strict_ordered_ratio", "unordered_recall", "unordered_precision", "unordered_f1", "overall_score"]
+    )
+
+    if not workspace.exists() or not workspace.is_dir():
+        return {k: 0.0 for k in ALL_CRITERIA} | {"error": f"workspace not found: {workspace}"}
+
+    pred_files = sorted(
+        [p for p in workspace.glob("*.tex") if p.stem.isdigit()],
+        key=lambda p: int(p.stem),
+    )
+    pred_contents = [normalize_tex(f.read_text(encoding="utf-8")) for f in pred_files]
+    num_pred = len(pred_contents)
+    scores = {}
+
+    scores["files_created"] = 1.0 if num_pred > 0 else 0.0
+
+    for i in range(1, num_gt + 1):
+        key = f"ordered_match_{i}"
+        if i - 1 < num_pred and i - 1 < num_gt:
+            scores[key] = 1.0 if pred_contents[i - 1] == gt_contents[i - 1] else 0.0
+        else:
+            scores[key] = 0.0
+
+    ordered_correct = 0
+    for i in range(min(num_pred, num_gt)):
+        if pred_contents[i] == gt_contents[i]:
+            ordered_correct += 1
+    scores["strict_ordered_ratio"] = round(ordered_correct / num_gt, 4)
+
+    gt_matched = set()
+    pred_matched = set()
+    for pi, pc in enumerate(pred_contents):
+        for gi, gc in enumerate(gt_contents):
+            if gi not in gt_matched and pc == gc:
+                gt_matched.add(gi)
+                pred_matched.add(pi)
+                break
+
+    recall = len(gt_matched) / num_gt if num_gt > 0 else 0.0
+    precision = len(pred_matched) / num_pred if num_pred > 0 else 0.0
+    f1 = (2 * precision * recall / (precision + recall)) if (precision + recall) > 0 else 0.0
+
+    scores["unordered_recall"] = round(recall, 4)
+    scores["unordered_precision"] = round(precision, 4)
+    scores["unordered_f1"] = round(f1, 4)
+
+    scores["overall_score"] = round(
+        0.7 * scores["strict_ordered_ratio"] + 0.3 * scores["unordered_f1"], 4
+    )
+
+    return scores

01_Productivity_Flow_task_2_table_tex_download/tests/grader.py

@@ -0,0 +1,76 @@
+"""Execute an embedded WildClawBench grader as a Harbor verifier."""
+
+from __future__ import annotations
+
+import json
+import math
+import runpy
+import traceback
+from pathlib import Path
+from typing import Any
+
+from transcript_loader import load_transcript, write_openclaw_compat_transcript
+
+
+WORKSPACE = "/tmp_workspace"
+TESTS_DIR = Path("/tests")
+VERIFIER_LOG_DIR = Path("/logs/verifier")
+
+
+def _numeric_rewards(scores: dict[str, Any]) -> dict[str, float]:
+    rewards: dict[str, float] = {}
+    for key, value in scores.items():
+        if isinstance(value, bool):
+            rewards[str(key)] = float(value)
+        elif isinstance(value, (int, float)) and math.isfinite(float(value)):
+            rewards[str(key)] = float(value)
+
+    if "overall_score" not in rewards:
+        numeric_values = list(rewards.values())
+        rewards["overall_score"] = (
+            sum(numeric_values) / len(numeric_values) if numeric_values else 0.0
+        )
+    return rewards
+
+
+def _write_json(path: Path, value: Any) -> None:
+    path.write_text(
+        json.dumps(value, indent=2, ensure_ascii=False, default=str) + "\n",
+        encoding="utf-8",
+    )
+
+
+def main() -> int:
+    VERIFIER_LOG_DIR.mkdir(parents=True, exist_ok=True)
+    try:
+        namespace = runpy.run_path(str(TESTS_DIR / "checks.py"))
+        grade = namespace.get("grade")
+        if not callable(grade):
+            raise TypeError("/tests/checks.py must define a callable grade()")
+
+        transcript = load_transcript()
+        write_openclaw_compat_transcript(transcript)
+        scores = grade(
+            transcript=transcript,
+            workspace_path=WORKSPACE,
+        )
+        if not isinstance(scores, dict):
+            raise TypeError(
+                f"WildClawBench grade() returned {type(scores).__name__}, expected dict"
+            )
+
+        _write_json(VERIFIER_LOG_DIR / "wildclaw_score.json", scores)
+        _write_json(VERIFIER_LOG_DIR / "reward.json", _numeric_rewards(scores))
+        return 0
+    except Exception as exc:
+        traceback.print_exc()
+        _write_json(
+            VERIFIER_LOG_DIR / "wildclaw_score.json",
+            {"overall_score": 0.0, "error": str(exc)},
+        )
+        _write_json(VERIFIER_LOG_DIR / "reward.json", {"overall_score": 0.0})
+        return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

01_Productivity_Flow_task_2_table_tex_download/tests/gt/1.tex

@@ -0,0 +1,39 @@
+\begin{table}[h!]
+    \centering
+    \resizebox{\textwidth}{!}{
+        \begin{tabular}{l|cccccc}
+        \toprule
+        \multirow{2}{*}{\textbf{Model}} & \multicolumn{5}{c}{\textbf{Video Categories}} & \multirow{2}{*}{\textbf{Overall}} \\
+        & Live-action & Animation  & Stock & YouTube & Shorts & \\\midrule
+        \multicolumn{7}{l}{\textit{Proprietary models}} \\
+        GPT-4V \cite{gpt4v} & 34.8/39.2/31.3 & 27.4/31.9/24.0 & 40.7/\underline{46.7}/36.1 & 33.8/40.1/29.2 & 34.8/46.1/28.0 & 34.4/40.8/29.7 \\
+        GPT-4o \cite{gpt4o} & 39.8/\underline{42.1}/37.8 & 35.8/39.1/33.1 & 44.0/46.6/41.7 & 35.9/\underline{41.5}/31.7 & 39.9/47.9/34.2 & 39.2/\underline{43.4}/35.7 \\
+        Gemini-1.5-Flash \cite{geminiteam2024gemini15unlockingmultimodal} & 34.8/36.4/33.3 & 29.2/32.5/26.5 & 39.4/39.7/39.1 & 34.3/38.6/30.9 & 35.6/42.4/30.7 & 34.8/37.9/32.1 \\
+        Gemini-1.5-Pro \cite{geminiteam2024gemini15unlockingmultimodal} & 36.4/36.4/36.4 & 30.7/31.8/29.7 & 42.2/40.7/43.8 & 34.0/36.7/31.6 & 37.0/42.4/32.7 & 36.2/37.6/34.8 \\
+        \midrule
+        \multicolumn{7}{l}{\textit{Open-source models ($>$10B)}} \\
+        PLLaVA-34B \cite{xu2024pllava} & 29.3/34.9/25.2 & 20.9/32.0/15.6 & 35.1/42.5/29.9 & 28.9/40.8/22.3 & 25.6/41.9/18.4 & 28.2/38.4/22.3 \\
+        VideoLLaMA2-72B \cite{cheng2024videollama2} & 27.3/29.3/25.6 & 19.7/21.7/18.1 & 33.9/37.0/31.3 & 27.7/33.0/23.8 & 26.5/33.1/22.1 & 27.1/30.8/24.2 \\
+        LLaVA-OV-72B \cite{li2024llavanext} & 31.7/32.8/30.7 & 27.7/30.6/25.2 & 38.0/39.6/36.6 & 34.1/34.7/33.5 & 33.8/41.8/28.4 & 33.2/35.9/30.9 \\
+        LLaVA-Video-72B \cite{zhang2024video} & 33.5/36.3/31.1 & 28.6/31.7/26.1 & 39.3/41.1/37.6 & 32.8/34.7/31.1 & 35.7/42.8/30.6 & 34.0/37.3/31.3\\
+        Qwen2-VL-72B \cite{qwen2vl} & 32.1/33.7/30.6 & 27.6/32.6/23.9 & 41.1/41.2/41.1 & 32.0/38.1/27.7 & 32.1/41.0/26.4 & 33.2/37.3/29.9\\
+        InternVL2.5-78B \cite{chen2024expanding} & 25.3/31.5/21.1 & 21.8/28.8/17.6 & 33.5/38.1/29.9 & 31.0/38.5/25.9 & 31.1/41.7/24.8 & 28.6/35.7/23.9\\
+        Tarsier-34B \cite{wang2024tarsierrecipestrainingevaluating} & 38.5/39.6/37.5 & 32.2/35.8/29.2 & 41.7/46.4/37.8 & 34.5/41.1/29.7 & 34.0/44.1/27.7 & 36.3/41.4/32.4  \\
+        % VILA-40B \\
+        \midrule
+        \multicolumn{7}{l}{\textit{Open-source models ($<$10B)}} \\
+        Video-LLaVA-7B \cite{lin2023video} & 19.4/24.3/16.2 & 15.3/21.2/11.9 & 27.0/33.5/22.7 & 21.2/31.9/15.8 & 18.5/29.4/13.5 & 20.4/28.1/16.0 \\
+        VideoLLaMA2-7B \cite{cheng2024videollama2} & 25.1/28.7/22.2 & 20.4/25.5/17.0 & 32.6/35.5/30.2 & 27.5/33.5/23.4 & 24.5/34.1/19.2 & 26.2/31.5/22.4 \\
+        LLaVA-OV-7B \cite{li2024llavanext} & 31.2/33.2/29.3 & 26.8/29.0/25.0 & 38.1/39.1/37.1 & 30.6/32.1/29.2 & 31.4/38.3/26.6 & 31.7/34.3/29.4 \\
+        LLaVA-Video-7B \cite{zhang2024video} & 31.4/35.2/28.4 & 27.6/32.9/23.8 & 36.7/39.7/34.1 & 33.0/\textbf{39.5}/28.3 & 33.4/42.5/27.5 & 32.5/37.9/28.4 \\
+        Qwen2-VL-7B \cite{qwen2vl} & 27.7/32.5/24.2 & 22.2/28.0/18.4 & 37.0/36.1/38.0 & 30.7/35.5/27.0 & 29.1/37.6/23.8 & 29.6/33.9/26.3 \\ 
+        InternVL2.5-8B \cite{chen2024expanding} & 26.6/32.0/22.8 & 21.3/28.9/16.9 & 32.7/37.2/29.1 & 27.9/35.4/23.0 & 28.9/39.9/22.7 & 27.6/34.7/22.9 \\
+        Tarsier-7B \cite{wang2024tarsierrecipestrainingevaluating} & 36.6/38.5/34.8 & 29.3/34.6/25.5 & 39.6/44.7/35.5 & 33.0/39.2/28.4 & 33.6/44.6/26.9 & 34.6/40.3/30.2 \\
+        \midrule
+        Tarsier2-7B & \underline{\textbf{44.4}}/\textbf{41.9}/\underline{\textbf{47.3}} & \underline{\textbf{39.3}}/\underline{\textbf{39.5}}/\underline{\textbf{39.1}} & \underline{\textbf{45.7}}/\textbf{45.4}/\underline{\textbf{46.0}} & \underline{\textbf{36.0}}/38.4/\underline{\textbf{33.9}} & \underline{\textbf{43.7}}/\underline{\textbf{48.9}}/\underline{\textbf{39.4}} & \underline{\textbf{42.0}}/\textbf{42.8}/\underline{\textbf{41.1}} \\
+        \bottomrule
+        \end{tabular}
+    }
+    \caption{Evaluation results on DREAM-1K. We report F1/Precision/Recall scores for each category and for the overall dataset. For open-source models, all results are tested with their official checkpoint and inference code under recommended setting. SOTA results of comparable scale ($<$10B) are bolded and overall best results are underlined.}
+    \label{tab:dream-1k}
+    \end{table}

01_Productivity_Flow_task_2_table_tex_download/tests/gt/10.tex

@@ -0,0 +1,19 @@
+\begin{table}[t]
+    \centering
+    \scriptsize
+    \setlength{\tabcolsep}{3pt} % 调整列间距
+    \resizebox{\textwidth}{!}{\begin{tabular}{l|ccc|cc|c}
+    \toprule
+    \multirow{2}{*}{\textbf{Model}} & \multicolumn{3}{c|}{\textbf{Caption}} & \multicolumn{2}{c|}{\textbf{Video QA}} & \multirow{2}{*}{\textbf{Hallucination}} \\
+    & DREAM-1K & TempCompass-cg & Vinoground-Text & Short & Long & \\
+    \midrule
+    Tarsier2-7B & 42.0 & 66.6 & 65.8 & 56.1 & 62.8 & 74.0  \\
+    \midrule
+    \quad \textit{w/o DPO} & 40.8 ($\downarrow$1.2) & 62.1 ($\downarrow$6.5) & 60.6 ($\downarrow$5.6) & 56.2 ($\uparrow$0.1) & 63.2 ($\uparrow$0.4) & 71.9 ($\downarrow$2.1) \\
+    \quad \textit{w/o NS} & 41.5 ($\downarrow$0.5) & 61.1 ($\downarrow$5.5) & 59.8 ($\downarrow$6.0)& 56.1 ($\downarrow$0.0)  & 62.8 ($\downarrow$0.0) & 72.9 ($\downarrow$1.1) \\
+    \quad \textit{w/o PF} & 40.5 ($\downarrow$1.5) & 65.1 ($\downarrow$1.5) & 67.6 ($\uparrow$1.8) & 56.0 ($\downarrow$0.1) & 62.3 ($\downarrow$0.5) & 74.2 ($\uparrow$0.2) \\
+    \bottomrule
+    \end{tabular}}
+    \caption{Ablation study for DPO training phase, negative sampling (NS) and preference data filtering (PF) strategies.}
+    \label{tab:dpo_ablation}
+\end{table}

01_Productivity_Flow_task_2_table_tex_download/tests/gt/11.tex

@@ -0,0 +1,24 @@
+\begin{table}[t]
+    \centering
+    \scriptsize
+    \setlength{\tabcolsep}{3pt} % 调整列间距
+    \resizebox{\textwidth}{!}{\begin{tabular}{l|ccc|cc|c}
+    \toprule
+    \multirow{2}{*}{\textbf{Model}} & \multicolumn{3}{c|}{\textbf{Caption}} & \multicolumn{2}{c|}{\textbf{Video QA}} & \multirow{2}{*}{\textbf{Hallucination}} \\
+    & DREAM-1K & TempCompass-cg & Vinoground-Text & Short & Long & \\
+    \midrule
+    % Tarsier2-7B & 42.0 & 66.6 & 65.8 & 56.1 & 62.6 & 74.0  \\
+    % \midrule
+    % Tarsier-7B & 34.6 & 55.3 & 29.8 & 45.6 & 46.3 & 56.3 \\
+    % \makecell[l]{+ \textit{Recaption FT}} & 31.6 & 52.9 & 27.4 & 44.1 & 45.2 & 46.3 \\
+    % \makecell[l]{+ \textit{Original FT}} & xxx & xxx & xxx & xxx & xxx & xxx \\
+    % \midrule
+    Qwen2-VL-7B \cite{qwen2vl} & 31.2 & 54.2 & 40.0 & 49.4 & 60.3 & 51.9 \\
+    \midrule
+    \makecell[l]{+ \textit{Original FT}} & 35.2 ($\uparrow$4.0) & 49.9 ($\downarrow$4.3) & 39.0 ($\downarrow$1.0) & 46.9 ($\downarrow$2.5) & 55.4 ($\downarrow$4.9) & 43.0 ($\downarrow$8.9) \\
+    \makecell[l]{+ \textit{Recaption FT}} & 39.5  ($\uparrow$8.3) & 67.7 ($\uparrow$13.5) & 55.0 ($\uparrow$15.0) & 52.5 ($\uparrow$3.1) & 56.8 ($\downarrow$3.5) & 68.5 ($\uparrow$16.6) \\
+    \bottomrule
+    \end{tabular}}
+    \caption{The experimental results of recaptioning. ``\textit{Recaption FT}'' represents fine-tune the model on the Tarsier2-Recap-585K dataset. ``\textit{Original FT}'' represents fine-tune the model with the same videos as Tarsier2-Recap-585K but taking their original labels as target output.}
+    \label{tab:recaption}
+\end{table}

01_Productivity_Flow_task_2_table_tex_download/tests/gt/12.tex

@@ -0,0 +1,28 @@
+\begin{table}[h!]
+    \centering
+    \resizebox{\textwidth}{!}{
+    \begin{tabular}{l |c c c c}
+         \toprule
+         \textbf{Configuration}   & \textbf{Pre-training} & \textbf{SFT-1} & \textbf{SFT-2} & \textbf{DPO} \\\midrule
+         VLM init.                & Qwen2-VL-7B & Tarsier2-Pre-trian & Tarsier2-SFT-1 & Tarsier2-SFT-2 \\
+         Optimizer name           & \multicolumn{4}{c}{AdamW} \\
+         Optimizer $\beta_1$      & \multicolumn{4}{c}{$0.9$}\\
+         Optimizer $\beta_2$      & \multicolumn{4}{c}{$0.999$}\\
+         Optimizer eps            & \multicolumn{4}{c}{$1e^{-6}$}\\
+         Learning rate            & $2e^{-5}$ & $2e^{-5}$ & $2e^{-6}$ & $1e^{-6}$\\
+         Learning rate schedule   & \multicolumn{4}{c}{cosine} \\
+         Training steps           & 200,000 & 5,000 & 5,000 & 1,000\\
+         Warm-up steps            & 1,000 & 250 & 250 & 100 \\
+         Weight decay             & \multicolumn{4}{c}{0.01}\\
+         Gradient clip            & \multicolumn{4}{c}{1.0} \\
+         Dropout rate             & \multicolumn{4}{c}{0.0}\\
+         Global batch size        & 384 & 64 & 64 & 64 \\
+         Max pixels         & \multicolumn{4}{c}{460,800} \\
+         Frames per video         & [8,128] & 16 & 16 & 16 \\
+         Numerical precision      & \multicolumn{4}{c}{bfloat16} \\
+         \bottomrule
+    \end{tabular}
+    }
+    \caption{Training hyper-parameters of \modelname}
+    \label{tab:hyperparam}
+\end{table}

01_Productivity_Flow_task_2_table_tex_download/tests/gt/13.tex

@@ -0,0 +1,109 @@
+\begin{table}[h!]
+    \centering
+    \small
+    \setlength{\tabcolsep}{3pt} % 调整列间距
+    \resizebox{\textwidth}{!}{
+    \begin{tabular}{llll}
+    \toprule
+    \multicolumn{4}{l}{\textit{\textbf{Video Captioning}}} \\
+    WebVid~\cite{bain2021frozen} (2.9M) & 
+    LSMDC~\cite{rohrbach2017movie} (109K) &
+    TGIF~\cite{li2016tgif} (105K) & 
+    ActivityNet~\cite{krishna2017dense} (38K) \\
+    Charades~\cite{sigurdsson2016hollywood} (16K) & 
+    Charades-Ego~\cite{sigurdsson2018charades} (6K) & 
+    YouCook2~\cite{zhou2018youcook2} (9K) &
+    TACoS~\cite{regneri2013grounding} (18K)\\
+    Ego4D~\cite{grauman2022ego4d} (1.1M) &
+    Spoken Moments~\cite{monfort2021spoken} (493K) & 
+    Multi-Moments~\cite{monfort2021multi} (997K) &
+    TREC-VTT~\cite{awad2023trecvid} (64K) \\
+    ShareGPT-4o-video~\cite{sharegpt4o} (2K) &
+    MovieStory101\cite{he2024storyteller} (11K)  &
+    GPT4o-labeled Caption$^\dagger$ (2.5M) &
+    Human-labeled Caption$^\dagger$ (145K) \\
+    Film\&TV Commentary$^\dagger$ (11.5M) & 
+    \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Action Recognition}}} \\
+    HMDB~\cite{kuehne2011hmdb} (5.8K) & 
+    COIN~\cite{tang2019coin} (10K) & 
+    SSV2~\cite{goyal2017something} (169K) &
+    Kinetics-700~\cite{carreira2017quo} (537K) \\
+    FineAction~\cite{liu2022fineaction} (82K) & 
+    RareAct~\cite{miech2020rareact} (2K) & 
+    20BN-jester~\cite{materzynska2019jester} (46K) & \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Video QA}}} \\
+    CLEVRER~\cite{yi2019clevrer} (83K) & 
+    TGIF-QA~\cite{jang2017tgif} (72K) & 
+    EgoQA~\cite{fan2019egovqa} (5K) &
+    VideoInstruct~\cite{maaz2023video} (89K) \\
+    LLaVA-Video-178K~\cite{zhang2024video} (165K) & 
+    M4-Instruct-video~\cite{li2024llava} (255K) & 
+    GPT4o-labeled QA$^\dagger$ (16.2K) & 
+    \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Grounding}}} \\
+    DiDeMo~\cite{anne2017localizing} (82K) & 
+    AVA~\cite{gu2018ava} (28K) & 
+    E.T. Instruct 164K~\cite{liu2024etbench} (147K) & 
+    Object Tracking$^\dagger$ (745K) \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Video Self-Supervised Training}}} \\
+    Frame Order Prediction$^\dagger$ (825K) \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Intent Recognition}}} \\
+    Oops!~\cite{epstein2020oops} (15K) & & & \\
+    
+    
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Multi-Image Understanding}}} \\
+    VIST~\cite{huang2016visual} (38K) & 
+    MMDU~\cite{liu2024mmdu} (45K) & 
+    M4-Instruct-image~\cite{li2024llava} (616K) & 
+    Image Retrival$^\dagger$ (533K) \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Single-Image Understanding}}} \\
+    ShareGPT4V~\cite{chen2023sharegpt4v} (95K) &
+    LLaVA-1.5~\cite{liu2023improved} (643K) &
+    ShareGPT-4o-image\cite{sharegpt4o} (57K) & 
+    MS COCO~\cite{lin2014microsoft} (566K) \\
+    Flicker~\cite{plummer2015flickr30k} (145K) & 
+    LLaVA-ReCap-CC3M~\cite{li2024llava} (2.9M) & 
+    Visual Genome~\cite{krishna2017visual} (759K) & 
+    SBU Captions~\cite{ordonez2011im2text} (860K) \\
+    GPT4o-labeled Caption$^\dagger$ (1.13M) \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Image OCR}}} \\
+    RCTW-17~\cite{shi2017icdar2017} (8K) & 
+    LSVT~\cite{sun2019icdar} (430K) & 
+    ReCTS~\cite{zhang2019icdar} (20K) & 
+    Art~\cite{bhagavatula2019abductive} (5.6K) \\
+    COCOTextV2~\cite{veit2016coco} (16K) & 
+    CORD-v2~\cite{park2019cord} (1K) & 
+    HierText~\cite{long2022towards} (10K) & 
+    MSRA-TD500~\cite{yao2012detecting} (465) \\
+    IC03~\cite{lucas2005icdar} (499) & 
+    SynthDoG-en~\cite{kim2022donut} (100K) & 
+    SynthDoG-zh~\cite{kim2022donut} (100K) & \\
+    
+    \midrule
+    \multicolumn{4}{l}{\textit{\textbf{Text Generation}}} \\
+    OpenOrca~\cite{lian2023openorca} (995K) &
+    ShareGPT~\cite{vicuna2023} (80K) & & \\
+    
+    \bottomrule
+    \end{tabular}
+    }
+    \caption{Datasets and their sizes used in \modelname pre-training. $\dagger$ indicates in-house datasets.}
+    \label{tab:pretraining-datasets}
+    \end{table}

01_Productivity_Flow_task_2_table_tex_download/tests/gt/14.tex

@@ -0,0 +1,28 @@
+\begin{table}[h!]
+    \centering
+    \resizebox{0.9\textwidth}{!}{%
+    \begin{tabular}{cl|ccc}
+    \toprule
+        \textbf{Capability} & \textbf{Benchmark} & \textbf{Tarsier1-7B} & \textbf{Tarsier1-7B-Qwen} & \textbf{Tarsier2-7B} \\
+        
+    \midrule
+    \multirow{3}{*}{Caption} & DREAM-1K & 34.6/30.2/40.3 & 38.4/40.6/36.4 & 40.8/42.5/39.3 \\
+    & TempCompass-cg & 55.3 & 59.3 & 60.1 \\
+    & Vinoground-Text & 29.8 & 48.6 & 60.2 \\
+    \midrule
+    \multirow{3}{*}{Video QA Short} & MVBench & 62.6 & 69.8 & 72.8  \\
+    & TVBench & 45.8 & 51.0 & 53.5 \\
+    & TOMATO & 28.6 & 36.5 & 39.5 \\
+    \midrule
+    \multirow{3}{*}{Video QA Long} & Video-MME & 42.2 & 58.9 & 65.3 \\
+    & LongVideoBench & 39.8 & 52.1 & 58.3 \\
+    & TemporalBench & 56.9 & 61.9 & 68.7 \\
+    \midrule
+    \multirow{2}{*}{Hallucination} & EventHallusion-Y/N & 70.9 & 75.6 & 77.8 \\
+    & EventHallusion-Desc & 41.6 & 48.6 & 49.1\\
+    \bottomrule
+    \end{tabular}
+    }
+    \caption{Detailed results of the ablation study for pre-training. For the captioning task, results are reported after the SFT stage. For other tasks, results are reported after the pre-training stage. }
+    \label{tab:appendix-pretrain_detailed_results}
+\end{table}

01_Productivity_Flow_task_2_table_tex_download/tests/gt/15.tex

@@ -0,0 +1,27 @@
+\begin{table}[h!]
+    \centering
+    \resizebox{0.9\textwidth}{!}{%
+    \begin{tabular}{cl|ccc}
+    \toprule
+        \textbf{Capability} & \textbf{Benchmark} & \makecell[c]{\\ pre-train} & \makecell[c]{\textbf{Tarsier2-7B}\\ SFT w/o grounding} & \makecell[c]{\\ SFT} \\
+    \midrule
+    \multirow{3}{*}{Caption} & DREAM-1K & 35.2/36.8/33.7 & 37.4/38.6/36.3  & 40.8/42.5/39.3 \\
+    & TempCompass-cg & 50.5 & 50.2 & 60.1 \\
+    & Vinoground-Text & 57.2 & 60.6 & 60.2 \\
+    \midrule
+    \multirow{3}{*}{Video QA Short} & MVBench & 72.8 & 71.9 & 72.5  \\
+    & TVBench & 53.5 & 54.5 & 54.2 \\
+    & TOMATO & 39.5 & 41.3 & 41.9 \\
+    \midrule
+    \multirow{3}{*}{Video QA Long} & Video-MME & 65.3 & 64.0 & 64.7 \\
+    & LongVideoBench & 58.3 & 54.7 & 58.2 \\
+    & TemporalBench & 68.7 & 66.9 & 66.6 \\
+    \midrule
+    \multirow{2}{*}{Hallucination} & EventHallusion-Y/N & 77.8 & 80.1 & 84.4 \\
+    & EventHallusion-Desc & 49.1 & 56.2 & 59.4 \\
+    \bottomrule
+    \end{tabular}
+    }
+    \caption{Detailed results of the ablation study for SFT.}
+    \label{tab:appendix-sft_detailed_results}
+\end{table}

01_Productivity_Flow_task_2_table_tex_download/tests/gt/16.tex

@@ -0,0 +1,27 @@
+\begin{table}[h!]
+    \centering
+    \resizebox{0.9\textwidth}{!}{%
+    \begin{tabular}{cl|cccc}
+    \toprule
+        \textbf{Capability} & \textbf{Benchmark} & \textbf{Tarsier2-7B} & \textit{w/o DPO} & \textit{w/o NS} & \textit{w/o PF}\\
+    \midrule
+    \multirow{3}{*}{Caption} & DREAM-1K & 42.0/42.8/41.1 & 40.8/42.5/39.3 & 41.5/44.5/39.0 & 40.5/39.9/41.1 \\
+    & TempCompass-cg & 66.6 & 60.1 & 62.1 & 65.1 \\
+    & Vinoground-Text & 65.8 & 60.2 & 60.6 & 67.6 \\
+    \midrule
+    \multirow{3}{*}{Video QA Short} & MVBench & 71.5 & 72.5 & 72.2 & 71.7 \\
+    & TVBench & 54.7 & 54.2 & 54.9 & 54.6 \\
+    & TOMATO & 42.0 & 41.9 & 41.3 & 41.8 \\
+    \midrule
+    \multirow{3}{*}{Video QA Long} & Video-MME & 64.5 & 64.7 & 64.3 & 64.4 \\
+    & LongVideoBench & 58.6 & 58.2 & 58.6 & 57.4 \\
+    & TemporalBench  & 65.3 & 66.6 & 65.4 & 65.2 \\
+    \midrule
+    \multirow{2}{*}{Hallucination} & EventHallusion-Y/N & 84.6 & 84.4 & 85.1 & 84.8 \\
+    & EventHallusion-Desc & 63.3 & 59.4 & 60.7 & 63.5 \\
+    \bottomrule
+    \end{tabular}
+    }
+    \caption{Detailed results of the ablation study for DPO.}
+    \label{tab:appendix-dpo_detailed_results}
+\end{table}