fix(tools): 隐藏 output_path 以修复 LLM 绕过 ARTIFACT_ROOT 的持久化缺陷

- cadquery_tool/lux3d_tool schema: 明确指示 LLM 不要传 output_path，
描述改为推荐省略以使用配置的 artifact 目录
- prompts.py SYSTEM: 在系统提示里再次强调 do NOT pass output_path

根因：LLM 按旧 schema 默认值提示，每次主动传相对路径 outputs/xxx.step，
safe_path 拼到 WORKDIR=/app 而非 ARTIFACT_ROOT=/data/outputs，
导致生成物全部写到容器临时目录，HF 持久化卷始终为空。

agent_core/prompts.py

@@ -12,14 +12,14 @@ For CadQuery tasks:
 - Write CadQuery Python code.
 - The code must assign the final model to a variable named result.
 - You may use cadquery as cq; it is pre-imported by the tool.
-- Call execute_cadquery with the code and optional output_path.
+- Call execute_cadquery with just the code; do NOT pass output_path. The tool writes outputs into the configured artifact directory under a unique run directory automatically.
 - Each successful tool call writes into a unique run directory to avoid overwriting previous models.
 - If execute_cadquery returns ok=false, inspect the structured error, fix the code, and call execute_cadquery again.
 
 For image-to-3D tasks:
 - The user must provide an image path from the workspace or configured artifact directory.
 - If the user asks for image-to-3D generation but does not provide an image path, ask for the image path. Do not guess.
-- Call generate_3d_model with the image_path and optional output_path.
+- Call generate_3d_model with just the image_path; do NOT pass output_path. The tool writes outputs into the configured artifact directory under a unique run directory automatically.
 - Do not ask the user to read files, convert images to base64, call APIs, or download results manually.
 
 When a tool returns ok=true, report output_path, run_dir, and manifest_path to the user.

agent_core/tools/cadquery_tool.py

@@ -259,8 +259,7 @@ TOOL_SCHEMA = {
         Execute CadQuery Python code and export the result as a STEP file.
         The code must assign the final CadQuery model to a variable named result.
         cadquery is pre-imported as cq.
-        output_path is optional. If it is a .step/.stp file path, the file name is used inside a unique run directory.
-        If it is a directory path, model.step is written inside a unique run directory under that root.
+        Do NOT supply output_path in normal usage. The tool automatically writes files to the configured artifact directory under a unique run directory. Only set output_path if the user explicitly requests a different location.
     """).strip(),
     "input_schema": {
         "type": "object",
@@ -271,7 +270,7 @@ TOOL_SCHEMA = {
             },
             "output_path": {
                 "type": "string",
-                "description": "Optional workspace-relative STEP file path or output directory. Defaults to outputs/model.step.",
+                "description": "Advanced: custom STEP file path or output directory. Omit this field to use the configured artifact directory (recommended).",
             },
         },
         "required": ["code"],

agent_core/tools/lux3d_tool.py

@@ -297,7 +297,7 @@ TOOL_SCHEMA = {
             },
             "output_path": {
                 "type": "string",
-                "description": "Optional workspace-relative model file path or output directory. Defaults to outputs.",
+                "description": "Advanced: custom model file path or output directory. Omit this field to use the configured artifact directory (recommended).",
             },
         },
         "required": ["image_path"],