ia-brainstorming
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user requests and existing codebase files (modules, docs, commits) to ground its brainstorming process. While it incorporates multiple human-in-the-loop confirmation gates (Phase 1 summary, Phase 2.5 synthesis, and Phase 4 approval) to mitigate risks, the ingestion of external content constitutes an attack surface.
- Ingestion points: User idea descriptions and codebase file content in Phase 0 and Phase 1.
- Boundary markers: The process requires explicit summaries and user confirmation steps before moving from understanding to approach exploration and finally to document creation.
- Capability inventory: The skill is authorized to read files, read git commits, and write new design documents to the
docs/brainstorms/directory. - Sanitization: The instructions do not specify any explicit escaping or sanitization of external content before it is interpolated into the agent's context.
Audit Metadata