The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill retrieves comment data from the vendor's official API at https://redfox.hk/story/api/youtube/videoComments. This communication is essential for the skill's primary function and targets the author's own infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a data processor for untrusted third-party content (YouTube comments).
Ingestion points: Comments are pulled from the YouTube API through the redfox.hk endpoint within scripts/youtube_comment_search.py.
Boundary markers: While no specific prompt boundary markers are used, the skill is instructional in nature and treats the data as objects for analysis rather than command input.
Capability inventory: The skill possesses network access (via urllib) and local file-writing capabilities (to generate HTML reports in ~/Downloads/QoderReports).
Sanitization: The escape_html function in scripts/youtube_comment_search.py is used to sanitize comment content, mitigating risks when generating the analysis report.
[CREDENTIALS_UNSAFE]: The skill follows secure practices by instructing users to manage the REDFOX_API_KEY via environment variables rather than hardcoding credentials in scripts or instructions.

youtube-comment