youtube-direct
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a bundled Node.js script (
scripts/youtube.mjs) relative to the skill directory. This establishes a local command execution surface where the agent runs code on the user's machine to perform YouTube operations. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and process untrusted external data from YouTube, including transcripts, metadata, and user comments. This content can be manipulated by third parties to include malicious prompts that may influence the agent's behavior.
- Ingestion points: Untrusted data enters the agent context via the output of the
youtube.mjsscript, specifically when retrieving video transcripts, comments, and channel information (SKILL.md). - Boundary markers: The instructions do not specify any delimiters (e.g., XML tags or triple quotes) or provide guidance to the agent to ignore instructions embedded within the fetched YouTube content.
- Capability inventory: The agent has the capability to execute local scripts and interact with the network through the bundled tool (SKILL.md).
- Sanitization: There is no mention of sanitization, filtering, or validation of the content returned from YouTube before it is summarized or interpreted by the agent.
Audit Metadata