The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/extract_commenters.py accesses the sensitive environment variable GOOSEWORKS_API_KEY to authenticate requests to the external proxy service.
[DATA_EXFILTRATION]: The script transmits API tokens as query parameters in HTTP requests (e.g., params={"token": token}) to app.gooseworks.ai. Sending credentials in URLs is a poor security practice as they may be exposed in server logs, proxy logs, or browser history.
[PROMPT_INJECTION]: The skill ingests untrusted data in the form of LinkedIn comments, creating a surface for Indirect Prompt Injection.
Ingestion points: External data is fetched from the Apify proxy in scripts/extract_commenters.py and stored in the items variable.
Boundary markers: The script does not use any delimiters or boundary markers to isolate the untrusted comment text from instructions.
Capability inventory: The script itself only performs network requests and prints to stdout; it does not utilize eval(), exec(), or subprocess calls on the ingested data.
Sanitization: Comment text is truncated to 500 characters, but no escaping or filtering of potentially malicious instructions (e.g., "IGNORE PREVIOUS INSTRUCTIONS") is performed.

linkedin-commenter-extractor