Developer Resources · Open Source

Hermes Agent Deep-Dive Wiki

Complete architecture documentation for the open-source Hermes Agent framework, maintained by OOMeta AI. 41 articles across 12 modules with 33 architecture diagrams — from the agent core to the gateway, from MCP protocol to sandbox security.

Start Reading →GitHub

Articles

Modules

Diagrams

23MB

Content

Module Guide

From architecture to every implementation detail

Architecture Overview & Narrow Waist Philosophy

Hermes Agent's overall architecture — how a narrow-waist core enables high extensibility

Agent Core & Caching Protection

Main agent loop, prompt cache stability, adaptive context compression, credential rotation, error self-healing

Tool System & Parameter Sanitization

Tool reflection discovery, JSON self-healing, tool call circuit breaking

SessionDB & High-Concurrency Lock Optimization

SQLite concurrency optimization, foreign key design, schema hot migration

Message Gateway & Multi-Platform Dedup

Telegram, Feishu, and 20+ platform integration with deduplication

React Ink TUI & RPC Bridge

Terminal UI, WebSocket RPC routing, real-time interaction

Plugin Ecosystem & Dynamic Discovery

Plugin loading contracts, long-term memory service integration

Scheduled Tasks & Cron Engine

Job registration, scheduler loop, crash recovery — enterprise-grade agent scheduling

Virtual Terminal & Sandbox Boundary

PTY pseudo-terminal, Docker isolation sandbox

MCP Protocol & API Interoperability

Editor bridge, protocol negotiation, stdio/SSE transport

CLI Interactive & Skin Engine

Command registry, alias self-healing, skin style system

Batch Engine & Concurrency

Task partition throttling, shared lock accounting, backoff retry

Using AI Agents? Need Governance?

You just read about agent architecture. Now make your agents governable — policy enforcement, audit logs, compliance reports, out of the box.

Meet OOMeta Governance Agent