testing
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION] (SAFE): The skill uses strong imperative language to enforce a workflow for quality control. It also processes external codebase data, creating a standard indirect prompt injection surface. Evidence: 1. Ingestion: Codebase and project memory. 2. Boundary markers: Absent. 3. Capability: Command execution (tests) and file modification. 4. Sanitization: Absent.
- [COMMAND_EXECUTION] (SAFE): The workflow includes running test suites to verify functionality. This is the intended primary purpose of a testing skill and is considered safe behavior in this context.
- [DATA_EXFILTRATION] (SAFE): The skill interacts with local project memory and context indexes via provided interfaces. No evidence of unauthorized network access or secret exfiltration was found.
Audit Metadata