codebase-understanding
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and analysis of external repositories, creating a surface for indirect prompt injection if those codebases contain adversarial content.
- Ingestion points: The skill instructions direct the agent to read repository files, manifests, and documentation using tools like ripgrep and find (SKILL.md, reconnaissance.md).
- Boundary markers: The workflow mandates a classification system (VERIFIED, INFERRED, UNKNOWN) for all claims, which serves as a cognitive boundary to distinguish evidence from potentially malicious injected content (SKILL.md).
- Capability inventory: The skill utilizes filesystem navigation, search tools, and repository mapping capabilities to explore the target codebase.
- Sanitization: No explicit automated sanitization or filtering of the content retrieved from external codebases is defined, relying instead on the agent's internal safety protocols and the evidence-based classification workflow.
Audit Metadata