youtube-full
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- Persistence Mechanisms (HIGH): The skill documentation indicates that the setup script
tapi-auth.jssaves the API key to the user's shell profile. Modifying startup files like.bashrcor.zshrcis a persistent behavior that can be used for malicious access. - Indirect Prompt Injection (LOW): The skill fetches video transcripts and metadata from YouTube which are untrusted external sources.
- Ingestion points: YouTube transcript and metadata retrieved via
curlfromtranscriptapi.com. - Boundary markers: None mentioned; external data is likely interpolated directly into the agent context.
- Capability inventory: Ability to execute Node.js scripts and network requests via
curl. - Sanitization: No sanitization or filtering of external content is specified.
- Command Execution (MEDIUM): The skill invokes local Node.js scripts (
tapi-auth.js) during the registration and verification phases. The actual source of these scripts is not provided for analysis, hiding the specific logic used to modify system files.
Recommendations
- AI detected serious security threats
Audit Metadata