codebase-design
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a 'Design It Twice' workflow in 'references/design-it-twice.md' that spawns multiple parallel sub-agents to generate competing interface designs.
- Ingestion points: The skill ingests user-provided class definitions, module interfaces, and design constraints to serve as the context for sub-agent technical briefs.
- Capability inventory: The skill utilizes the 'Agent' tool to dynamically trigger multiple sub-agent executions based on its analysis of the user's codebase.
- Boundary markers: Although the instructions suggest creating independent briefs for sub-agents, there are no explicit boundary markers or instructions to disregard potential commands embedded within the user's code or architectural descriptions.
- Sanitization: The skill does not implement sanitization or filtering of the user-provided data before passing it into the prompts used to guide the sub-agents.
Audit Metadata