The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's instructions in SKILL.md direct the agent to execute Python scripts (fetch_trending.py and write_trending_note.py) directly via the command line to process data.
[EXTERNAL_DOWNLOADS]: The fetch_trending.py script performs network requests to github.com. It includes a fallback mechanism in the fetch_html function that disables SSL certificate validation (ssl.CERT_NONE) if an initial connection error occurs. This bypasses standard security protocols and makes the connection vulnerable to Man-in-the-Middle (MitM) attacks in untrusted network environments.
[PROMPT_INJECTION]: The skill ingests untrusted content (repository names and descriptions) from the GitHub Trending page and writes it into markdown files. While it uses strip_tags for HTML and basic escaping for markdown tables, it does not sanitize the text for natural language instructions, creating an indirect prompt injection surface if the agent later processes these notes.

github-trending