minimax-music-3
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of natural language guidelines and formatting rules for prompt engineering. It does not perform any programmatic actions other than directing the agent to save output to a specific project subdirectory.
- [SAFE]: The frontmatter includes
disable-model-invocation: true, which is a security restriction that limits the agent's ability to execute tools, thereby reducing the potential attack surface. - [SAFE]: No instances of prompt injection, obfuscation, data exfiltration, or persistence mechanisms were detected in the instructions or metadata.
Audit Metadata