talking-head-podcast-recut
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, prompt injections, or obfuscation were detected. The skill is entirely instructional and provides a structured production workflow with heavy emphasis on verification and safety triggers.
- [INDIRECT_PROMPT_INJECTION]: While the skill is designed to ingest and process external content (recordings and transcripts), it includes robust defensive instructions. It mandates a "context test" to prevent misrepresentation and requires a verbatim layer for all quotes to ensure accuracy. The risk of the agent being misled by instructions embedded in the data is mitigated by these explicit verification steps. The severity is internally assessed as SAFE.
- [EXTERNAL_DOWNLOADS]: The skill contains numerous references to official documentation and industry standards (e.g., FTC endorsement guides, W3C accessibility guidelines, EBU loudness standards, and official platform help pages for YouTube, TikTok, and LinkedIn). These references are used for informational purposes to guide the agent's behavior and compliance checks. All identified domains are well-known and reputable services.
Audit Metadata