The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill involves fetching video metadata (such as titles and descriptions) from external, attacker-controllable websites via yt-dlp --dump-json. This creates a surface for indirect prompt injection where malicious instructions could be embedded in the metadata.
Ingestion points: Metadata output from the yt-dlp utility as described in SKILL.md.
Boundary markers: No delimiters or warnings are provided to help the agent distinguish between data and instructions.
Capability inventory: The skill uses shell command execution to run yt-dlp and writes files to the local system.
Sanitization: The skill does not provide instructions for sanitizing or filtering the retrieved metadata before the agent processes it.
[COMMAND_EXECUTION]: The skill relies on shell commands to perform its core tasks. While the documentation correctly recommends using double quotes for URL placeholders to mitigate risk, the overall security depends on the agent's implementation of input sanitization for user-provided URLs.

video-download