伐谋 - 任务定义与评估器生成
PassAudited by ClawScan on May 1, 2026.
Overview
This instruction-only skill is coherent for creating FaMou task artifacts, but users should expect it to read project files, overwrite named outputs, and run generated Python for validation.
This looks reasonable for a FaMou artifact generator. Before installing or using it, be comfortable with the agent reading the target project directory, overwriting problem.md/evaluator.py/init.py/prompt.md, and running the generated Python validation command. Use a clean workspace or version control for important projects.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Existing task files may be changed or overwritten during normal use.
The skill can replace an existing task-definition file. This is expected for an artifact generator, but it still mutates local project state.
若 `problem.md` 已存在，合并后完整覆盖写入。
Use version control or ask the agent to show diffs before overwriting existing files.
Generated scripts could read or write local files or consume compute resources if the generated logic is broad or flawed.
The validation workflow executes local Python code, including generated evaluator/init artifacts. This is purpose-aligned, but users should understand that code will run locally.
运行评估器验证 `init.py`: ```bash python evaluator.py <path_to_init.py> ```
Run it in the intended project directory, review generated Python before execution when the task is sensitive, and consider a disposable workspace for experiments.
Project data or configuration details may be incorporated into the generated task documents and prompts.
The skill uses local project files and configuration as task context. That is useful and purpose-aligned, but such files may contain sensitive details or misleading content.
检查数据文件、代码脚本、README、配置，能从中回答的问题不再提问。
Avoid running it over directories containing unrelated secrets, and review generated files for sensitive information before sharing them.
Complex datasets may be processed by an additional skill beyond this one.
The skill may delegate complex data analysis to another skill, which could receive project data or context; the boundary for that delegation is not described here.
如果数据复杂，调用 famou-data-analysis 技能 辅助。
Confirm what data will be shared with the helper skill, especially for private or regulated datasets.