Hermes Agent Complete Guide

Open-source, self-hosted AI Agent with long-term memory and Skills system

GitHub: NousResearch/hermes-agentOfficial Site: hermes-agent.nousresearch.comChinese Community: hermesagent.org.cn

📖 Introduction

Hermes Agent is an open-source AI Agent developed by Nous Research, designed for long-term task execution and continuous operation scenarios. Unlike IDE assistants, Hermes emphasizes:

• Cross-session memory - Remembers your projects, preferences, and work habits long-term
• Reusable Skills - Distill solved problems into reusable skills
• Multi-platform messaging gateways - Stay online via Telegram, Discord, Feishu, etc.
• Self-hosted - Full control over data and runtime environment

✨ Core Features

1. Long-term Memory & Skills

Cross-session memory

• Remembers project context, code structure, personal preferences
• Continuously learns your work habits
• Every conversation builds on historical context

Skills system

• Distill solved problems into reusable Skills
• Compatible with agentskills.io open format
• Supports team sharing and community contributions

2. MCP & Tool Integration

40+ built-in tools

• Terminal command execution
• File system operations
• Browser automation
• Image processing
• TTS (text-to-speech)
• Multi-model inference

MCP support

• Compatible with Model Context Protocol
• Extensible custom tools
• Flexible toolset configuration

3. Multi-platform Messaging Gateways

Social platforms

• Telegram
• Discord
• Slack
• WhatsApp
• Signal
• WeChat

Business platforms

• Enterprise WeChat
• Feishu (Lark)
• DingTalk

Use case: Keep the Agent online across multiple platforms, receiving tasks and pushing results anytime, anywhere.

4. Automated Scheduling

Built-in Cron

• Auto-generate daily reports
• Scheduled backups
• System health checks
• Timed reminders
• Information scraping

5. Model Compatibility

Chinese models

• Qwen (Tongyi Qianwen)
• GLM (Zhipu)
• Kimi (Moonshot)
• MiniMax

International models

• Claude (Anthropic)
• Gemini (Google)
• Codex (OpenAI)

Other interfaces

• OpenRouter and other relay services
• OpenAI-compatible interfaces
• Local model support

🚀 Installation

System Requirements

• Linux / macOS / WSL2 / Windows
• Network connection (for downloads and model calls)
• Optional: Docker environment

Quick Install

Linux / macOS / WSL2

curl -fsSL https://res1.hermesagent.org.cn/install.sh | bash

Windows PowerShell

irm https://res1.hermesagent.org.cn/install.ps1 | iex

Note: WSL2 is the recommended long-term solution for most Windows users.

Configure Model

# Start setup wizard
hermes setup

# Configure model
hermes model

Start Conversation

# Start full TUI
hermes

After launch, includes:

• Multi-line input
• Command completion
• Context compression
• Tool output streaming
• Session history

Connect Messaging Gateway

# Configure messaging gateway
hermes gateway setup

# Start gateway
hermes gateway

📚 Use Cases

Case 1: Terminal Task Execution

# Interact with Hermes directly in terminal
hermes

# Example: Ask Hermes to analyze project code
> Analyze this project's code structure and find potential optimizations

# Example: Generate daily report
> Generate a daily report based on today's git commits

Case 2: Multi-platform Continuous Work

Configure Telegram Bot

hermes gateway setup telegram

Use case

• Send tasks to Hermes via Telegram
• Hermes processes and pushes results
• Interact with Agent anytime, anywhere

Case 3: Automated Tasks

Create scheduled tasks

# Edit crontab
hermes cron edit

# Example: Generate daily report at 9am
0 9 * * * hermes task daily-report

# Example: System health check every hour
0 * * * * hermes task system-check

Case 4: Long-term Project Assistant

Project initialization

# Create project config
hermes project init my-project

# Set project context
hermes project config --name "My Project" --tech "Python, React"

Continuous collaboration

• Hermes remembers project structure and code conventions
• Every conversation builds on project context
• Skills distill project-specific knowledge

🔧 Advanced Configuration

Environment Isolation

# Create isolated environments
hermes profile create work
hermes profile create personal

# Switch environment
hermes profile switch work

Command Approval

# Configure commands requiring approval
hermes config set approval.required "rm,git push,docker"

# Set approval timeout
hermes config set approval.timeout 300

Container Isolation

# Run tools in isolated container
hermes config set sandbox.enabled true
hermes config set sandbox.image "hermes-sandbox:latest"

Vision Features

# Enable visual understanding
hermes config set vision.enabled true

# Use vision feature
> Analyze UI issues in this screenshot [upload image]

🆚 Hermes Agent vs OpenClaw

Feature	Hermes Agent	OpenClaw
Token consumption	Lower (~30%)	Higher
Process transparency	High, execution steps visible	Medium
Long-term memory	Native support	Limited
Skills system	Built-in, compatible with agentskills.io	Needs extra config
Messaging gateways	Multi-platform native support	Needs extra integration
Migration support	hermes claw migrate one-click	-
Model support	Broad Chinese & international models	Primarily Claude
Community	Active Chinese community	English community

Selection advice:

• Need long-term memory and Skills → Hermes Agent
• Need multi-platform messaging gateways → Hermes Agent
• Need lower Token consumption → Hermes Agent
• Primarily use Claude → Both work

🛠️ Migrating from OpenClaw

# One-click migration command
hermes claw migrate

# Migration includes:
# - Config files
# - Historical sessions
# - Custom settings

📊 Best Practices

1. Project Initialization Flow

# 1. Create project directory
mkdir my-project && cd my-project

# 2. Initialize Hermes project
hermes project init

# 3. Configure project info
hermes project config --name "My Project" --description "Project description"

# 4. Set tech stack
hermes project config --tech "Python, FastAPI, PostgreSQL"

# 5. Start collaborating
hermes

2. Skills Development Flow

# 1. Create Skill
hermes skill create my-skill

# 2. Define Skill functionality
# Edit ~/.hermes/skills/my-skill/skill.yaml

# 3. Test Skill
hermes skill test my-skill

# 4. Publish Skill (optional)
hermes skill publish my-skill

3. Team Collaboration Config

# 1. Export project config
hermes project export > project-config.yaml

# 2. Team members import
git clone <project-repo>
cd project
hermes project import project-config.yaml

# 3. Share Skills
hermes skill share --team

🔗 Related Resources

• GitHub: NousResearch/hermes-agent
• Official Docs: hermes-agent.nousresearch.com
• Chinese Community: hermesagent.org.cn
• Nous Research: nousresearch.com

🆘 FAQ

Q: How is Hermes Agent different from IDE assistants?

A: Hermes emphasizes:

• Long-term context retention (cross-session memory)
• Reusable Skills distillation
• Multi-platform messaging gateways
• Self-hosting and data control

Q: Where should Chinese users start?

A:

1. Visit the Chinese community at hermesagent.org.cn
2. Check the Chinese installation guide
3. Windows users should prioritize the Windows installation guide
4. Join WeChat/Feishu community groups

Q: What deployment methods are supported?

A:

• Local computer
• VPS server
• Docker container
• SSH remote environment
• Cloud dev environments (GitHub Codespaces etc.)

Q: How to reduce Token consumption?

A:

• Enable context compression
• Use appropriate models (Chinese models cost less)
• Distill Skills to reduce repeated inference
• Configure command approval to avoid accidental operations

📝 Summary

Hermes Agent is an open-source AI Agent designed for long-term tasks and continuous operation, especially suitable for:

• ✅ Projects needing long-term memory and context retention
• ✅ Scenarios needing multi-platform messaging gateways
• ✅ Teams wanting to distill reusable Skills
• ✅ Scenarios concerned about Token consumption and cost
• ✅ Users needing self-hosting and data control

Get started now:

curl -fsSL https://res1.hermesagent.org.cn/install.sh | bash
hermes setup
hermes

---

License: MIT License · 2026 Nous Research