Podcast Analytics Expert
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- Indirect Prompt Injection (HIGH): The skill is vulnerable to indirect prompt injection due to its core functionality of processing external data combined with powerful tool access.
- Evidence Chain:
- Ingestion points: The skill fetches data from
api.transistor.fm,op3.dev, andapi.appstoreconnect.apple.comas seen inSKILL.md. - Boundary markers: No markers or 'ignore' instructions are used when handling API responses.
- Capability inventory: The skill metadata explicitly allows
Bash,Edit, andWebFetchtools. - Sanitization: No sanitization or schema validation is performed on the incoming data before potential processing.
- External Downloads (LOW): The skill performs network requests using
curland Pythonrequeststo non-whitelisted domains (transistor.fm,op3.dev,appstoreconnect.apple.com). While these are aligned with the stated purpose, they represent external data ingestion.
Recommendations
- AI detected serious security threats
Audit Metadata