test
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed to provide a read-only pass/fail verdict for test suites without applying fixes or modifications to the source code.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data including test reports and project configuration files, representing a standard attack surface.
- Ingestion points: Reads summary reports generated by the
testersubagent and project configuration fromdocs/project-config.json. - Boundary markers: Implements a strict 'READ-ONLY' constraint and delegates execution to an isolated subagent to prevent unintended side effects.
- Capability inventory: Includes file reading (configuration and reports), subagent spawning (via
tester), and file writing (appending to goal iteration logs). - Sanitization: Explicitly instructs the agent to never copy sensitive fixture data into the goal file, acting as a manual sanitization step for persistent logs.
- [COMMAND_EXECUTION]: The skill contains explicit prohibitions against direct command execution (e.g., 'NEVER invoke test commands directly'), requiring all execution to flow through the platform's
testersubagent.
Audit Metadata