architecture-refiner
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- SAFE (SAFE): No malicious patterns, obfuscation, or unauthorized access attempts were detected in the skill definition.
- Indirect Prompt Injection (LOW): The skill is designed to audit and refine external codebases, representing an ingestion surface for untrusted data. Evidence: [1] Ingestion points: Code analyzed during the Audit and Refinement phases. [2] Boundary markers: Absent. [3] Capability inventory: No executable scripts or system-access tools are provided in this skill. [4] Sanitization: None. Because the skill lacks executable capabilities, the risk remains minimal.
Audit Metadata