🤖 Hermes Agent
Hermes Agent 0.10 is here

The agent that grows with you.

Hermes is a strong fit if you want one agent that can keep long-running context, grow operator habits over time, and move from CLI to messaging without rebuilding your whole workflow.

Get Started ❯ First 30 Minutes

Quick Install

Before You Install

System Requirements

A minimal environment check before you run the installer.

Operating System

Supported environments for the current install flow.

macOSLinuxWindows WSL2Termux

Memory

Keep at least 2GB RAM available for a smooth setup and runtime baseline.

2GB+ RAM

Network

Internet access is required to download resources and call your LLM API provider.

Internet RequiredLLM API Access

Dependencies

Leave version placeholders if your stack depends on Python or Node.

Python: <version placeholder>Node: <version placeholder>
# Works on macOS, WSL2, Linux and Termux
$curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

# And then start the wizard!
$hermes setup
Installation Quickstart Telegram Setup Migrate from OpenClaw
Zero-Friction Transfer

One-Click Migration

Bring your OpenClaw setup into Hermes with a single command. Hermes imports everything from ~/.openclaw automatically.

$ hermes claw migrate
API KeysMemorySkills
Saves about 10 minutes of manual configuration
Before
  • Find and copy API credentials by hand
  • Rebuild memory files and local state
  • Re-import skills one by one
After
  • Run hermes claw migrate once
  • Hermes pulls config from ~/.openclaw
  • Resume with keys, memory, and skills ready
Gateway Setup

Connect to Your World

Connecting

Use the Hermes gateway to bridge directly into chat surfaces you already live in.

Telegram
Configure first, then verify the bot is reachable
Gateway Flow
01

Get Bot Token

Create a bot with BotFather and copy the Bot Token.

02

Run Setup

Use the official setup wizard first, then start the configured gateway.

hermes gateway setup
03

Verify Bot Access

Start the gateway and message the bot to verify. If pairing is enabled, approve the code it shows.

hermes gateway
Discord
Configure first, then verify the bot is reachable
Gateway Flow
01

Get Bot Token

Create an application bot in Discord Developer Portal and copy the Bot Token.

02

Run Setup

Use the official setup wizard first, then start the configured gateway.

hermes gateway setup
03

Verify Bot Access

Start the gateway, invite the bot or DM it, and approve pairing only if your setup prompts for one.

hermes gateway
Provider Strategy

Smart Model Selection

Pick the provider profile that matches your deployment goal during hermes setup.

Option Positioning Best For Setup Snippet
Nous Portal
Best-fit default
 Most aligned Hermes experience Users who want the smoothest first-run configuration and tuned defaults. 
Provider: Nous Portal
Profile: Best-fit default
Use case: Turnkey Hermes setup
OpenRouter
Multi-model choice
 Flexible provider switchboard Users comparing multiple hosted models or routing between vendors. 
Provider: OpenRouter
Model: anthropic/claude-3.7-sonnet
Use case: Multi-model routing
Local LLM
Privacy-first
 Runs against your own local endpoint Users prioritizing privacy, offline control, or self-hosted inference. 
Provider: Local LLM
Endpoint: http://localhost:1234/v1
Use case: Private local inference
Capabilities

Capability Map

Persistent Memory

Primary cross-session memory lives in MEMORY.md and USER.md, with SQLite-backed session search kept as a separate retrieval layer.

SOUL.md Identity

SOUL.md defines the global identity of a Hermes instance, while project-specific instructions belong in AGENTS.md.

Skills System

Hermes can discover, install, and reuse skills as structured procedures instead of repeating the same setup work every session.

Messaging Gateway

Configure platform adapters with hermes gateway setup, then run hermes gateway to keep Hermes reachable from chat surfaces.

MCP Integration

Hermes connects to MCP servers as scoped tool providers, making external systems available without widening access unnecessarily.

Delegation & Parallel Work

Use child agents for fresh-context review, parallel research, or other tasks that benefit from isolated reasoning and tool use.

What Changed Recently

Latest Release

Hermes Agent v0.10.0

The current official release introduces the Nous Tool Gateway for paid Nous Portal subscribers, with web search, image generation, TTS, and browser automation through one subscription.

Source: GitHub Releases
Hermes Core

Tool Gateway Flow

The release adds subscription-based tool access, per-tool opt-in, and clearer integration with hermes tools and hermes status.

Source: GitHub Releases
Gateway API

Where to Double-Check

Use GitHub Releases and the docs when you want to confirm platform support, setup flow, or a release-specific capability.

Source: Official Docs
Demo

See Hermes In Action

A quick demo helps with orientation, but exact setup details still belong in the docs and release notes.

This video is a demo, not a complete feature contract. Use the docs and release notes when you need exact setup or support details.

Open release notes

Common Use Cases

01

Personal Agent on Telegram

Install Hermes locally or on a small VPS, run the gateway, and keep a private assistant reachable from any device.

02

OpenClaw Migration Path

Move SOUL, memory, skills, and provider settings from ~/.openclaw without rebuilding your setup by hand.

03

Identity + Project Rules

Use SOUL.md for global identity and AGENTS.md for repo-specific instructions so behavior stays coherent across sessions.

04

MCP for Real Tool Access

Connect Hermes to scoped MCP servers for filesystems, GitHub, databases, and internal APIs without widening tool access too early.

05

Delegation for Heavy Work

Use delegation when you need isolated child agents for research, review, or parallel workstreams that would otherwise flood context.

06

Long-Running Agent Workflow

Run Hermes as a persistent CLI or messaging agent, keep session search available, and verify changes against official releases and docs.

View all shoutouts →

Frequently Asked Questions

What is Hermes Agent? +

Hermes Agent is an open-source AI agent by Nous Research. The official docs describe persistent memory, SOUL-based identity, messaging gateway support, MCP integration, and delegation for parallel work.

How can I run Hermes? +

Official installation supports macOS, Linux, Windows via WSL2, and Android through Termux. Native Windows is not the supported install path.

Which Large Language Models does it support? +

Hermes supports Nous Portal, OpenRouter, OpenAI, local endpoints, and other compatible providers described in the setup and model documentation.

How do I connect Telegram or Discord? +

Use hermes gateway setup to write the platform configuration, then run hermes gateway. Telegram and Discord each have dedicated setup guides in the official docs.

Official References

If you want the original docs, release notes, and repository pages, start here.

Repo Releases Docs