design-it-twice
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing content from external codebase files which could contain adversarial instructions.\n
- Ingestion points: The skill ingests codebase content via the Read, Grep, and Glob tools to provide context for design problems as specified in SKILL.md.\n
- Boundary markers: There are no explicit delimiters or instructions provided to the agent to distinguish between the provided codebase context and its internal design procedures.\n
- Capability inventory: The agent is granted access to the Task tool, creating a capability surface that could be targeted if malicious instructions were embedded in the read files.\n
- Sanitization: No input validation or sanitization is performed on the codebase content before it is used by the agent in its design reasoning process.
Audit Metadata