Open Source AI Agent · From Nous Research

The Complete Guide to Hermes Agent

Deploy Hermes once on your server. It learns your projects, builds its own skills, and stays reachable on every device you use — quietly getting smarter every day.

GitHub Stars
157K
Built-in Tools
40+
Platforms
22+
Stable Version
v0.14.0

Just a chat. Lots of jobs.

One persistent agent on your server — long-term memory, real browser control, scheduled jobs, and a self-evolving Skills system.

RUNS ON A TIMER

Tell It Once, It Runs Forever

Describe a recurring job once. Hermes runs it on schedule, survives restarts.

hermes · Slack
Post the daily standup prompt to #team at 9am every weekday
Scheduled — I'll post the prompt every weekday at 09:00 and save the replies to memory.
Skip US holidays, and ping me if nobody replies by 10
Linked your US holidays calendar — won't run on holidays. Added a 10:00 silence check that DMs you if fewer than 3 people replied. Next post: Mon Mar 24.
Also summarize the replies and post a recap to #leadership at 11
Done. The 11:00 follow-up clusters replies into 'in progress / blocked / done' and posts the recap to #leadership. Want a dry-run before it goes live tomorrow?
LEARNS YOU

Knows How You Work

Teach it your conventions once. Every future task ships in your voice.

hermes · Telegram
We use Conventional Commits — always prefix with feat:, fix:, chore:, etc.
Got it. Logged to your project preferences — I'll prefix every commit title with the right type from now on.
Also: every PR description needs a 'Risk' section at the end
Added. Every PR description I draft will end with a 'Risk' section listing what could break in prod, plus a rollback plan.
Draft the commit + PR for the auth bugfix I just merged
Commit: fix(auth): fall back to silent reauth on expired refresh token PR draft pushed to #98. Risk section flags the session migration as the highest-blast-radius change — rollback = revert PR #95.
DRIVES A BROWSER

Actually Opens The Page

Hermes drives a real browser — the page itself, not a search snippet.

hermes · Discord
What's new on Anthropic's blog this week?
Pulled 5 posts from anthropic.com/news (Mar 17–23). Top read: 'Claude Opus 4.7 release' (Mar 18) — adds 1M token context and faster tool use.
Open that post and pull out the API changes
Three API changes worth knowing: • `extended_thinking` now defaults to on • new `prompt_caching: "auto"` option • `legacy_max_tokens` removed (deprecated since 4.5)
Cross-check the changelog — anything in our codebase that breaks?
Yes — 3 endpoints still use `legacy_max_tokens`. I opened a draft PR replacing them with the new `max_tokens` field. CI's green, ready for your review.
LONG-TERM MEMORY

Remembers Months Back

Stop re-explaining yourself. Hermes pulls up decisions and reasoning from chats months back.

hermes · Telegram
What did we land on for that caching layer we talked about last month?
Found it — Feb 14 Slack thread. Picked Redis to reuse the existing BullMQ instance. Logged in #architecture.
I'm building it now but I've forgotten the details — give me a checklist
Based on that thread + your last 3 similar PRs: 1. `ioredis` wrapper with TTL per key prefix 2. Write-through invalidation 3. DB fallback on Redis failure 4. Hit/miss metrics in Grafana
What was Alex's pushback exactly?
Two concerns: (1) no fallback when Redis is down; (2) cross-service invalidation. The fallback was left open — resolve before you ship.

Get Your 24/7 AI Assistant in 5 Minutes

Install, configure, and start using Hermes in one short pass.

1

Install

The bootstrap script sets up uv and Python 3.11, clones the repo, and finishes initial setup without asking for sudo.
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
2

Configure

Connect Nous Portal, OpenRouter, or your own endpoint. The guided setup flow walks through the essential model settings.
# Interactive setup wizardhermes setup# Or choose your modelhermes model
3

Start Chatting

Launch Hermes and you immediately get the full interactive CLI with tools, memory, and skills ready to work.
hermes
4

Add Messaging Platforms (Optional)

When you want Hermes online all the time, wire up Telegram, Discord, Slack, or WhatsApp and run the gateway as a system service.
# Interactive gateway setup wizardhermes gateway setup# Start the messaging gatewayhermes gateway# Install as a system servicehermes gateway install
5

Stay Current

Run the updater whenever you want the latest code, dependencies, and fixes pulled into the local runtime.
hermes update
Native Windows support is still experimental. For a steadier setup, install WSL2 first and run Hermes inside it.

Migrating from OpenClaw?
Done in 3 steps

Use hermes claw migrate for one-shot migration — configs are preserved. Backup, import, diagnose in one flow.

Read the migration guide
hermes-sh
~ hermes claw migrate --dry-run
[INFO] Scanning for legacy OpenClaw instances.
[INFO] Found 3 nodes. Starting data mapping.
[PROGRESS]100%
✓ SUCCESS: Migration completed in 1.2s
✓ All configs migrated
✓ All 3 agents imported

Best Tutorials & Guides

HermesAgent 101 aims to be the best Hermes Agent resource hub.

Contributions welcome — add resources, improve content, share your experience.

EN
Official Docs

Hermes Agent Official Site

The official Hermes Agent website for project overview, features, and latest updates.
Official DocsVisit
EN
Official Docs

Hermes Agent Documentation

The official documentation covering installation, quickstart, architecture, gateways, memory, skills, MCP, and more.
Official DocsVisit
EN
Official Docs

NousResearch/hermes-agent

The official GitHub repository for Hermes Agent, including source code, issues, releases, and project activity.
GitHubVisit
EN
Official Docs

Hermes Agent v0.8.0 Release

The current stable release page, useful for quickly checking recent features and behavior changes.
GitHub ReleasesVisit
EN
Official Docs

Weixin (WeChat) Gateway for Hermes Agent

Official guide on integrating Hermes Agent with personal WeChat via the iLink Bot API.
Official DocsVisit
EN
Official Docs

Architecture Overview

A developer-facing architecture guide focused on the agent runtime, prompt assembly, and the core execution loop.
Official DocsVisit
ZH
Tutorials

Hermes Agent Quick Start Guide

Covers Hermes Agent installation, Claude Pro integration, cross-session auto-learning, and comparison with OpenClaw.
BlockTempoVisit
EN
Tutorials

Getting Started with Hermes Agent in Under an Hour

A getting-started tutorial by Anthony Maio covering installation, messaging platform setup, persistent memory, and the skill system.
LinkedInVisit
EN
Videos

Hermes Agent Setup Tutorial — Agent on VPS in 8 Minutes

A ByteGrad video tutorial demonstrating how to quickly set up Hermes Agent on a VPS.
YouTubeVisit
ZH
Videos

Hermes Agent Local Deploy + Telegram Tutorial

A Chinese video tutorial by Lingdu covering Hermes Agent local deployment with Telegram integration and troubleshooting tips.
YouTubeVisit
EN
Videos

Hermes Agent Full Setup Tutorial (Gemma 4)

A video tutorial by Bart Slodyczka on setting up Hermes Agent from scratch with Gemma 4.
YouTubeVisit
ZH
Videos

From OpenClaw to Hermes Agent: Setup, Migration & Demo

A Chinese video by NiceKate AI demonstrating the full migration process from OpenClaw to Hermes Agent with hands-on configuration.
YouTubeVisit
ZH
Tools

MiniMax Platform × Hermes Agent Setup

Set up Hermes Agent on MiniMax platform with the MiniMax-M2.7 model for autonomous programming.
MiniMaxVisit
EN
Tools

Model Context Protocol

The official MCP site. Important if you want to understand the protocol layer behind Hermes tool integrations.
MCPVisit
EN
Tools

Hermes Agent on OpenRouter

Hermes Agent app page on OpenRouter, showing usage analytics and 200+ supported models.
OpenRouterVisit
EN
Tools

Browser Use × Hermes Agent Integration

How to integrate Browser Use cloud browser automation with Hermes Agent, including agent self-registration.
Browser UseVisit
EN
Cloud Deployment

Daytona

One of the common execution backends used with Hermes, useful when evaluating isolated cloud runtimes.
DaytonaVisit
EN
Cloud Deployment

Hermes Agent VPS Hosting — One-Click Docker

Hostinger VPS hosting for Hermes Agent with one-click Docker setup and always-on access.
HostingerVisit
EN
Community

Hermes Agent GitHub Star History

View Hermes Agent GitHub star growth trajectory and contributor activity.
Star HistoryVisit
EN
Community

Hermes Agent SourceForge Mirror

Hermes Agent mirror on SourceForge for alternative download access.
SourceForgeVisit
EN
Analysis

What Is Hermes Agent? The OpenClaw Alternative

A MindStudio analysis comparing Hermes and OpenClaw on design philosophy, learning loops, and use cases.
MindStudioVisit
EN
Analysis

Hermes Agent: Self-Improving AI with Persistent Memory

Explores how Hermes Agent maintains continuous memory across sessions and autonomously develops new capabilities.
Yuv.aiVisit
EN
Analysis

Hermes Agent — OpenClaw's Rival? Differences and Best Use Cases

A Turing Post analysis breaking down the key differences between Hermes Agent and OpenClaw with best use cases.
Turing PostVisit
EN
Analysis

Hermes Agent Overview

An overview by Jimmy Song covering the learning loop, skill creation, cross-session memory, and multi-platform support.
Jimmy SongVisit

Not a Copilot. Not a Chatbot.

Hermes Agent is an open-source autonomous AI agent from Nous Research. Not an IDE copilot or a thin chatbot wrapper, but a long-running agent that lives on your server, keeps what it learns, and becomes more capable over time.

Lives Where You Do

One agent running across Telegram, Discord, Slack, WhatsApp, Signal, email, and the terminal. Start on your phone, pick it up on your laptop.

Grows The Longer It Runs

Persistent memory and auto-generated skills work together. It learns your projects and remembers how it solved that bug last time.

Real Sandboxing

Five execution backends to pick from: local, Docker, SSH, Singularity, Modal. Container hardening and namespace isolation — keep it off your main machine if you want to.

Scheduled Automations

Describe a cron job in plain English — "send me a news briefing every morning at 8" — and it runs itself. Results land in whichever gateway you choose.

Parallel Sub-Agents

Spin up isolated sub-agents that each have their own conversation and terminal. Multi-step pipelines collapse into a single near-zero-context RPC call.

Full Web & Browser Control

Search, scrape, navigate, click, type, screenshot — all end to end. Layer on vision, image generation, text-to-speech, and multi-model reasoning as needed.

Frequently Asked Questions

Quick answers to the questions we get most often.

Stay Current on Hermes

Only the updates that are actually worth your time.