brainstorm
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the
WebSearchtool and user-provided$ARGUMENTSto generate its ideation reports. It has the capability to write these results to project files within theplans/directory (e.g.,plans/{plan-id}/brainstorm-opportunity-map.md). While the skill instructs the agent to be skeptical and require evidence (>80% confidence), the absence of explicit delimiters or sanitization for external content creates a surface for indirect prompt injection. - Ingestion points: External data from the
WebSearchtool and user input throughask user question tool. - Boundary markers: The skill does not define explicit markers or "ignore" instructions for content retrieved via web searches.
- Capability inventory: The skill is authorized to write and amend markdown files in the project's
plans/directory. - Sanitization: No explicit sanitization or validation of external tool output is defined in the instructions.
Audit Metadata