brainstorm
Pass
Audited by Gen Agent Trust Hub on Mar 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill incorporates untrusted user-provided content into agent instructions and external tool queries without using boundary markers or sanitization.
- Ingestion points: User requests and discovery answers described in SKILL.md.
- Boundary markers: Absent during the Synthesize phase and in the final prompt structure.
- Capability inventory: Access to WebSearch and sub-agent delegation (Agent/task).
- Sanitization: No input validation or filtering instructions are provided for processed user data.
Audit Metadata