ai-podcast-creation

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides templates that interpolate untrusted external data into prompts for LLM and TTS models.
  • Ingestion points: User-provided content is intended to be placed in placeholders like <your-document-content>, <host-lines>, and <guest-lines> in SKILL.md.
  • Capability inventory: The skill utilizes the belt CLI to call remote AI models (Claude, Kokoro TTS, AI Music).
  • Sanitization: There are no clear boundary markers or instructions to the model to ignore adversarial instructions that might be contained within the processed documents or script inputs.
  • [COMMAND_EXECUTION]: The skill is configured to use the belt command-line interface via Bash(belt *). This allows the agent to execute specific platform tools for media generation and processing. The frontmatter correctly limits the scope of shell execution to this specific tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 01:13 PM
Security Audit — agent-trust-hub — ai-podcast-creation