The Agent Skills Directory

Prompt Injection (HIGH): Vulnerable to Indirect Prompt Injection.
Ingestion points: The skill uses yt-dlp to fetch external transcripts, video titles, and descriptions.
Boundary markers: None present; the skill lacks delimiters or instructions to ignore embedded commands in the downloaded data.
Capability inventory: The skill allows Bash, Read, and Write tools, providing a powerful execution environment for injected payloads.
Sanitization: No sanitization or validation of the external content is performed before processing.
Command Execution (MEDIUM): The skill relies on several external binaries (yt-dlp, eyeD3, and the undocumented ytmp3). The use of an unknown command like ytmp3 without a verifiable source or installation path represents a risk, and passing unvalidated YouTube metadata as arguments to these shell commands could lead to traditional command injection.

youtube