clawpilot-config
Warn
Audited by Gen Agent Trust Hub on May 5, 2026
Risk Level: MEDIUMCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill instructions direct the agent to locate and inspect files containing sensitive credentials and authentication tokens.
- Target files include ~/.hermes/.env and ~/.clawai/runtimes/ccconnect.json.
- Specific sensitive fields mentioned include gateway token, API_SERVER_KEY, and management tokens. While required for the stated purpose, this provides the agent with access to raw secrets.
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface.
- Ingestion points: Local configuration files (.env, .json) are read by the agent.
- Boundary markers: Absent.
- Capability inventory: File reading capabilities.
- Sanitization: Absent. The agent processes untrusted configuration data without specific safety boundaries.
Audit Metadata