The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides multiple templates for executing yt-dlp and ffmpeg shell commands. These commands are used to download media, merge streams, and extract metadata. While standard for this tool, shell command execution is a high-capability operation.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from the internet, which represents a potential attack surface for indirect prompt injection.
Ingestion points: The skill uses yt-dlp --dump-json to retrieve video metadata (titles, descriptions, etc.) from external, attacker-controlled URLs (SKILL.md).
Boundary markers: Absent. There are no instructions or delimiters provided to ensure the agent ignores potentially malicious instructions embedded within the downloaded metadata.
Capability inventory: The skill allows local command execution, file system writes (downloads), and metadata parsing.
Sanitization: Absent. No filtering or validation logic is suggested for the external metadata before it is processed by the agent.

video-download