podcast-production
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFENO_CODEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [NO_CODE]: The analyzed skill consists entirely of markdown documentation, instructional prompts, and episode templates. It contains no executable scripts, binaries, or command-line operations.
- [EXTERNAL_DOWNLOADS]: The skill references an external Node.js package '@clawfu/mcp-skills' in the YAML frontmatter mcp-server configuration field. This package provides the environment for the skill but no malicious code is present in the skill definition itself.
- [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface by instructions that process untrusted user data, such as interview transcripts in the 'Transform an Interview' section. Evidence: 1. Ingestion points: Transcription input in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: None (the skill lacks tools or shell access). 4. Sanitization: Absent. The risk is assessed as safe because the agent has no dangerous capabilities to exploit via injection.
Audit Metadata