brainstorming
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill demonstrates an attack surface for indirect prompt injection due to its integration of external project data with high-privilege operations.\n- Ingestion points: The skill instructs the agent to read current project files, documentation, and recent commit messages as specified in the 'Understanding the idea' section of SKILL.md.\n- Boundary markers: There are no instructions to utilize delimiters or specific boundary markers to isolate untrusted project data from the agent's core instructions.\n- Capability inventory: The skill has the ability to write design documents to 'docs/plans/', perform git commits, and invoke powerful external skills like 'superpowers:using-git-worktrees' and 'superpowers:writing-plans'.\n- Sanitization: No sanitization, validation, or escaping mechanisms are defined for the project data ingested during the brainstorming process.
Audit Metadata