acestep
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted natural language data which could contain malicious instructions designed to influence the agent's behavior during the audio generation process.
- Ingestion points: The
tools/music_gen.pytool accepts untrusted input through the--promptand--lyrics(via subshell command or string) arguments inSKILL.md. - Boundary markers: There are no specific boundary markers or explicit instructions provided in the command examples to help the agent distinguish between its own operational instructions and the user-provided text content.
- Capability inventory: The skill performs network operations (communicating with the RunPod API using
RUNPOD_API_KEY) and file system writes (creating.mp3,.wav, or.flacfiles). - Sanitization: No evidence of input validation or sanitization of the strings is present before they are passed into the command-line execution string.
Audit Metadata