The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell scripts (scripts/yt-transcript) and system utilities (yt-dlp, curl, pandoc) to process media and web content. These operations are necessary for the skill's primary functionality of content conversion.
[EXTERNAL_DOWNLOADS]: The skill fetches data from external URLs (YouTube transcripts and web pages) using yt-dlp and curl. This behavior is the intended purpose of the skill and uses well-known industry tools.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external sources (YouTube captions and HTML) that could contain malicious instructions aimed at influencing the agent's behavior.
Ingestion points: Video transcripts retrieved via yt-dlp in scripts/yt-transcript and raw HTML retrieved via curl in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or 'ignore embedded instructions' warnings for the agent to use when interpolating the untrusted external data into its context.
Capability inventory: The agent has access to shell execution (via yt-dlp, curl, pandoc) and file system write access for outputting processed files.
Sanitization: The yt-transcript bash script uses awk to strip some tags and clean whitespace, but lacks comprehensive security sanitization to prevent adversarial instructions within the content from being interpreted by the agent.

yt-structure