brainstorming
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to review project state files and documentation, which serves as a potential surface for indirect prompt injection where untrusted data could attempt to influence the agent's reasoning process. However, the risk is mitigated by explicit instructions prohibiting code execution or system modification during the brainstorming phase.\n
- Ingestion points: Project state files, documentation, plans, and prior decisions as specified in the 'Understand the Current Context' section of SKILL.md.\n
- Boundary markers: The skill does not define specific delimiters for project data, relying instead on its internal process flow.\n
- Capability inventory: The skill facilitates text-based reasoning and documentation. It explicitly disallows implementation, coding, or behavior modification in its instructions.\n
- Sanitization: No specific sanitization or filtering of project content is mentioned.\n- [SAFE]: The skill follows secure principles by using 'Understanding Locks' and mandatory decision logs to ensure transparency and user control throughout the design process.
Audit Metadata