Summary

This repository provides a comprehensive guide for setting up, migrating, and optimizing the Hermes Agent for production environments. It targets engineers and power users, offering a clear path from initial installation to advanced configurations like LightRAG, extensive platform integrations, and autonomous skill creation, enabling cost-effective and robust AI agent deployments.

How It Works

The guide offers an end-to-end approach with runnable artifacts, including skills, configurations, and reference architectures. It emphasizes modularity, allowing users to select the most efficient setup path. Core components include a flexible input gateway, intelligent model routing, a robust approval layer, integrated tools, and advanced memory systems like LightRAG for graph-based knowledge retrieval. The architecture prioritizes observability and cost control, facilitating production-ready deployments.

Quick Start & Requirements

• Primary Install/Run: A one-command VPS bootstrap script (scripts/vps-bootstrap.sh) automates setup on Debian/Ubuntu. Local installation uses scripts/install.sh followed by hermes setup and hermes or hermes dashboard.
• Prerequisites: Linux/macOS/WSL2/Android (Termux), Python 3.11+, Git, and at least one LLM provider API key are required. Optional dependencies include Ollama for local embeddings and a Nous Portal subscription for the Tool Gateway.
• Links: Quickstart guide (docs/quickstart.md), Roadmap (ROADMAP.md), Changelog (CHANGELOG.md), Ecosystem (ECOSYSTEM.md).

Highlighted Details

• One-Command VPS Bootstrap: Automates installation of Hermes, Node.js, Caddy, UFW, fail2ban, systemd units, and skill symlinks on fresh Debian/Ubuntu servers.
• LightRAG: Implements graph RAG for advanced memory, enabling retrieval of connected knowledge and relationships beyond simple text similarity.
• Autonomous Skill Creation: Hermes can automatically offer to save complex workflows as reusable skills, enhancing self-improvement.
• Extensive Integrations: Supports numerous LLM providers (Anthropic, OpenAI, Bedrock, Azure, etc.) and platforms (Telegram, Discord, Slack, Teams, QQBot, etc.).
• Interactive Configuration: Guided setup wizards (hermes setup, hermes config set) and an Ink TUI (hermes --tui) simplify configuration.
• Reproducible Benchmarks: Performance and cost data for 12 models across 5 tasks are available in benchmarks/.

Maintenance & Community

Developed by Terp AI Labs. Links to ROADMAP.md and CHANGELOG.md are provided. No explicit community channels (e.g., Discord, Slack) are mentioned in the README.

Licensing & Compatibility

The specific open-source license is not stated in the provided README text. The guide focuses on enabling production deployments and includes migration support from OpenClaw.

Limitations & Caveats

Native Windows support is absent; WSL2 is required. Piping internet scripts to bash carries inherent security risks. The LightRAG REST API lacks built-in authentication and must be secured. Migration from OpenClaw does not transfer session transcripts, cron jobs, or plugin-specific data. Running local models demands substantial VRAM.