groove-work-brainstorm
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No malicious override instructions or safety bypass attempts were found. The skill follows a logical flow for project scoping and dialogue.
- [DATA_EXFILTRATION]: The skill operates on local project files and templates. There are no network requests, hardcoded credentials, or attempts to access sensitive system files like SSH keys or environment variables.
- [REMOTE_CODE_EXECUTION]: No external scripts are downloaded or executed. Tool usage is restricted to file operations and specified git commands via bash, which are relevant to the skill's purpose.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided topics and project metadata. While this represents an ingestion surface, the risk is addressed by the agent's instruction to ask clarifying questions and push back on scope creep. Evidence chain: 1. Ingestion points: .groove/index.md and user arguments. 2. Boundary markers: Instructions to verify scope via dialogue. 3. Capability inventory: File read/write and git commands. 4. Sanitization: Standard markdown output generation.
Audit Metadata