route-tester
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCREDENTIALS_UNSAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [CREDENTIALS_UNSAFE] (HIGH): The skill includes hardcoded credentials for a test user ('testuser'/'testpassword') and the MySQL root password ('password1') within the documentation and command examples.
- [COMMAND_EXECUTION] (HIGH): The instructions direct the agent to execute a local script ('scripts/test-auth-route.js') and perform direct database operations via 'docker exec' with root privileges.
- [DATA_EXFILTRATION] (HIGH): The skill exposes and encourages interaction with sensitive file paths including '.env' and 'config.ini' which are known to store application secrets and environment configurations.
- [INDIRECT_PROMPT_INJECTION] (LOW): 1. Ingestion points: Processed API response bodies from local or remote endpoints. 2. Boundary markers: None identified in the instructional patterns. 3. Capability inventory: node script execution, curl operations, docker exec database access. 4. Sanitization: No sanitization logic described for handling external API outputs.
Recommendations
- AI detected serious security threats
Audit Metadata