spec-brainstorm
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the local repository (codebase implementations, README, and existing specs) to generate design recommendations. This presents an attack surface for indirect prompt injection, where malicious instructions hidden in the codebase could attempt to manipulate the design output.
- Ingestion points: The skill reads AGENTS.md, README, and codebase implementations during the Orient and Research steps (SKILL.md).
- Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions found within the repository files.
- Capability inventory: The skill is strictly limited to writing a design.md file and providing conversational responses. It is explicitly forbidden from modifying other files, creating plans, or executing code (SKILL.md).
- Sanitization: No sanitization or validation of the ingested repository content is performed before processing.
Audit Metadata