hermes-agent-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill acts as a technical resource providing documentation and code examples for the Hermes Agent framework. The content is educational and does not contain executable malicious patterns.\n- [EXTERNAL_DOWNLOADS]: The documentation mentions the project's official repository (github.com/NousResearch/hermes-agent) for installation via git clone. This reference is legitimate and follows standard open-source practices for development documentation.\n- [DYNAMIC_EXECUTION]: The skill describes architectural components for sandboxed code execution and a plugin-based extension system. These are documented with security considerations such as resource limits, restricted import lists, and communication via Unix Domain Sockets.\n- [DATA_EXFILTRATION]: The skill documents messaging gateway integrations and memory persistence using local database files (~/.hermes/memory.db). These interactions are presented as core architectural features for intended agent functionality, with no signs of unauthorized data harvesting or transmission.
Audit Metadata