The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. It fetches content from external WeChat URLs (mp.weixin.qq.com) and provides this content to the note-creator skill to generate structured output.
Ingestion points: WeChat article content is fetched in tools/wechat_archiver.py and tools/wechat_archiver_v2.py via wechat2md.
Boundary markers: The skill uses minimal boundary markers when passing content to the next stage, such as instructions in templates/execution-flow.md stating '文章内容：见 article.md'.
Capability inventory: The skill has access to Bash, Write, Edit, and Skill tools, and executes subprocesses to run helper scripts.
Sanitization: While titles and URLs are sanitized/normalized, the actual article body is not sanitized before being processed by the agent or subsequent skills.
[COMMAND_EXECUTION]: The orchestration scripts tools/wechat_archiver.py and tools/batch_archiver.py use subprocess.run to execute other Python scripts. While these calls target internal skill components like wechat2md.py and use the system's Python executable, they represent the execution of code across script boundaries.

wechat-archiver