The AI Agent That Grows With You
An open-source autonomous AI agent that lives on your server, remembers what it learns, and gets more capable the longer it runs.
curl -fsSL hermes-agent.sh | bashChatbots talk. Hermes works.
Most AI tools give you answers. Hermes Agent gives you outcomes.
Up and running in five minutes
No complex configuration. One install command, connect your model, and deploy.
Install
One curl command. Works on Linux, macOS, and WSL2. No complex setup required.
Configure
Connect your preferred AI model, set up platform integrations, and define your agent's capabilities.
Deploy
Your agent lives on your server, learns from every interaction, and grows more capable over time.
Built for real autonomy
Not a wrapper around an API. A complete agent framework with persistent memory, self-improving skills, and 40+ tools.
Persistent Memory
Multi-level memory system that mimics procedural learning. Short-term tasks through inference, long-term via searchable Skill Documents.
Self-Improving Skills
Synthesizes complex tasks into permanent Skill Documents — a live knowledge graph that grows with every deployment.
40+ Built-in Tools
Web search, terminal, browser automation, vision, image generation, code execution, subagent delegation, and more.
Multi-Platform Gateway
Telegram, Discord, Slack, WhatsApp, Signal, Email, and CLI — all from a single gateway process.
Any Model, No Lock-in
Works with OpenRouter (200+ models), OpenAI, Nous Portal, or your own endpoint. Switching requires no code changes.
Flexible Deployment
Runs on a $5 VPS, a GPU cluster, or serverless. 6 terminal backends: Local, Docker, SSH, Daytona, Singularity, Modal.
Your server. Your data. Your agent.
Your Server
Runs anywhere — $5 VPS, GPU cluster, or serverless. 6 terminal backends supported. All data stays local.
Hermes Platform
ReAct reasoning loop, persistent memory, self-improving skills, subagent delegation, and multi-platform gateway.
AI Models
Works with 200+ models via OpenRouter, OpenAI, Nous Portal, or your own endpoint. No vendor lock-in.
Real workflows, real outcomes
Hermes Agent adapts to your specific workflows and gets better at them over time.
DevOps Automation
Monitor deployments, manage infrastructure, and automate incident response with persistent context.
Customer Support
Handle support tickets across platforms with memory of past interactions and learned solutions.
Research Assistant
Browse the web, analyze documents, and build a growing knowledge base of your domain.
Code Review
Review pull requests, suggest improvements, and learn your team's coding standards over time.
Data Pipeline
Orchestrate data workflows, monitor quality, and adapt processes based on learned patterns.
Content Creation
Generate, edit, and schedule content across platforms with your brand voice learned over time.
One agent, every platform
Deploy once and reach your team on Telegram, Discord, Slack, WhatsApp, Signal, Email, or CLI — all from a single gateway.
Your infrastructure,
your rules
Run Hermes Agent on your own servers. No data leaves your infrastructure. No telemetry, no tracking, no cloud lock-in.
- Full data sovereignty
- No usage limits or API costs
- 6 terminal backends
- Works offline with local models
Supported Backends
Loved by developers
“Hermes Agent replaced three separate automation tools for us. The memory system means it actually gets better at our specific workflows.”
“The fact that it runs on our own infrastructure and learns from every interaction makes it fundamentally different from any chatbot wrapper.”
“We deployed it on a $10 VPS and it handles our entire Telegram support channel. The skills it's built up over 3 months are incredible.”
active deployments worldwide
Your server. Your Hermes.
Five minutes.
Install Hermes Agent, connect your model, and deploy an AI agent that actually gets things done — and gets better every day.
Free & open-source · MIT License · No telemetry