helpmetest-self-heal
Pass
Audited by Gen Agent Trust Hub on Mar 10, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill is developed by the author help-me-test and utilizes corresponding vendor-specific tools (mcp__helpmetest-*), indicating a legitimate relationship between the skill's author and its functionality.
- [COMMAND_EXECUTION]: The skill performs command execution and file updates via helpmetest_run_interactive_command and helpmetest_upsert_test. These high-privilege operations are restricted to the testing environment and are necessary for the skill's primary function of repairing broken tests.
- [SAFE]: Persistence and background monitoring are implemented via standard agent features such as the /loop skill or background Tasks. These methods are used to handle test failure events in real-time and do not attempt to modify system startup configurations or user shell profiles.
- [SAFE]: A surface for indirect prompt injection exists because the skill processes external test error logs (ingestion point: event.error in the Monitoring Loop). While it lacks explicit boundary markers or sanitization for these logs, the risk is mitigated by the autonomous classification workflow and the limited scope of the vendor tools.
Audit Metadata