astartes-coding-custodes
Pass
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions are focused entirely on software engineering best practices, such as modularity, type safety, and error handling consistency. It does not request or perform any suspicious operations.- [PROMPT_INJECTION]: The skill operates on user-provided code and project files, which serves as an ingestion point for untrusted data (Indirect Prompt Injection surface). However, the mandated 'Strict Review' and 'Minimal Change Plan' phases provide structural defenses against unintended instruction following by requiring the agent to validate all changes against existing architecture.
- Ingestion points: Reads project source code and architecture files from the environment.
- Boundary markers: None explicitly defined.
- Capability inventory: Modifies local source files during the implementation phase.
- Sanitization: None present.
Audit Metadata