The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the uvx command (part of the uv Python toolchain) to execute yt-dlp for media processing. It also references standard system utilities like sed and jq for post-processing data.
[EXTERNAL_DOWNLOADS]: The skill fetches the yt-dlp package from the Python Package Index (PyPI) via uvx. It also facilitates the download of video, audio, and metadata content from YouTube servers to the local file system.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection.
Ingestion points: Untrusted content such as video titles, descriptions, and transcripts are ingested from YouTube into the agent's context (SKILL.md).
Boundary markers: While URLs are quoted in shell commands, there are no instructions to delimit or ignore instructions embedded within the fetched metadata or transcripts.
Capability inventory: The skill has the capability to write files to various local directories (SKILL.md).
Sanitization: The skill does not explicitly sanitize the external content it retrieves before it might be processed by subsequent agent steps.

youtube