The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/distill.py script uses os.system() to automatically install the requests library if it is not found. While this targets a legitimate library, programmatically executing shell commands is a risk factor.
[EXTERNAL_DOWNLOADS]: The skill makes several network requests to https://redfox.hk via scripts/gzh.py to fetch article UUIDs and full article content. These requests are authenticated using a user-supplied API key.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted article content from WeChat public accounts, creating a potential surface for indirect prompt injection attacks.
Ingestion points: External article data is fetched in scripts/gzh.py and saved to output/{博主名}_完整文章.json, which is intended for AI analysis.
Boundary markers: The skill lacks explicit boundary markers or instructions for the AI to ignore embedded commands within the processed articles.
Capability inventory: The skill possesses file-writing capabilities (output/ directory) and the ability to execute shell commands via os.system() in scripts/distill.py.
Sanitization: The clean_text function in scripts/distill.py performs basic HTML tag stripping and newline normalization, but does not sanitize the content against malicious prompt instructions.

investor-distiller