hermes-update-doctor
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill incorporates a comprehensive 'Untrusted Content Boundary' policy. It explicitly directs the agent to treat processed data (logs, repository files, web content) as evidence rather than instructions, effectively mitigating risks associated with indirect prompt injection.- [DATA_EXFILTRATION]: The skill logic enforces privacy by sanitizing remote URLs and summarizing process logs to remove credentials or private hostnames before they are included in the final report.- [COMMAND_EXECUTION]: The safety contract mandates that any system mutations, such as process management or repository writes, are strictly approval-gated and not performed during the diagnostic phase.- [SAFE]: No obfuscation, unauthorized network operations, or hardcoded credentials were found. The included validation and test scripts are limited to the Python standard library and serve legitimate development purposes.
Audit Metadata