The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/fetch_transcript.py uses subprocess.run to call the yt-dlp utility for metadata extraction. The command is constructed using a fixed list of arguments and a URL built from an 11-character video ID that is validated via regular expression, preventing arbitrary command injection.
[EXTERNAL_DOWNLOADS]: The skill depends on yt-dlp and youtube-transcript-api. During operation, it connects to YouTube's public infrastructure to retrieve video data. These are established services and the integration follows standard practices for the intended functionality.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted transcript data from YouTube.
Ingestion points: Transcript snippets are fetched via scripts/fetch_transcript.py and provided to the agent context for synthesis.
Boundary markers: While the skill uses structured templates, it lacks explicit delimiters or instructions to the agent to ignore any malicious prompts embedded within the transcript text itself.
Capability inventory: The skill can execute yt-dlp (via subprocess) and read local user-provided cookie files for authentication.
Sanitization: There is no evidence of filtering or sanitization of the transcript content before it is processed by the AI.

youtube-transcript-dossier