Hermes Agent Deployment Guide: Step-by-Step Instructions to Build Your First AI Assistant

Resource Sharing Tutorials & Documents
1

A Beginner’s Guide to Installing and Using Hermes Agent in 10 Minutes: Build Your Personal AI Assistant from Scratch — An AI Agent That Gets Smarter the More You Use It.

Video Tutorial: https://www.bilibili.com/video/BV1yQR8BhEhm/

Introduction

Have you ever had this experience? Every time you open an AI chat window, you have to re-explain your identity, preferences, and work background. Each time your Agent performs a task, it’s like its “first day on the job”—all prior experiences reset, and you’re forced to manually fix the same bugs over and over again.

What if there were an AI assistant that takes just minutes to deploy, remembers every conversation you’ve had, turns execution experience into reusable Skills, and stays active 24/7 on WeChat, Feishu, or Telegram ready to respond anytime? Such an agent doesn’t just complete “a single task”—it continuously learns your working style.

This is Hermes Agent.

An open-source, self-hosted AI Agent system, Hermes supports mainstream models such as Kimi, GLM, Claude, and Gemini. Its core design centers on “long-term memory + skill evolution”: it remembers past mistakes, builds up reusable skills, compresses context, and makes repeated tasks faster and more token-efficient over time. Compared to similar tools, Hermes offers more transparent tool calling and lower migration costs (migrate from OpenClaw with just one command), making it a truly long-term personal Agent system.

In this guide, I’ll walk you step-by-step through environment setup, model configuration, TUI conversation, and message gateway integration to complete your first Hermes Agent deployment.

Admin Discussion Group: 767557452

About Hermes Agent

Hermes Agent is a lightweight, open-source, self-hosted AI Agent with long-term memory and skill accumulation capabilities, optimized for Chinese users’ environments and needs. Its key advantages include:

  • :feather: Highly Flexible Deployment: Supports local PCs, VPS, Docker, WSL2, etc., across Linux/macOS/Windows, with domestic mirror acceleration for installation;
  • :brain: Learns and Evolves Over Time: Remembers your projects, preferences, and workflows across sessions, turning solved problems into reusable Skills that get smarter with use;
  • :hammer_and_wrench: Rich Toolset: Supports over 40 tools including MCP, terminal, file operations, browser, image processing, TTS, and automation via cron (e.g., daily reports, backups, system checks);
  • :electric_plug: Available on All Platforms: Integrates with QQ, WeChat, Feishu, DingTalk, Telegram, Discord, and more, for instant access anytime;
  • :puzzle_piece: Full Model Compatibility: Works with Qwen, GLM, Kimi, MiniMax, Claude, Gemini, OpenAI-compatible APIs, and local models—optimized for China’s network environment.

Official HermesAgent website: https://url.zeruns.com/HermesAgent

Preparation

I’ll be using a Linux-based cloud server for demonstration, but you can also deploy it on your Raspberry Pi, Mac, mini PC, or other devices. However, using a cloud server may be more convenient—e.g., if you want to create a website, you can instruct the AI to write and directly deploy a fully accessible site.

Recommended Cloud Providers:

I’ll use a RainCloud server for this tutorial. First, register a RainCloud account using the promo link or code (zeruns) below. After registration, log in to the dashboard, click Cloud Server → Purchase Cloud Server.

  • RainCloud Referral Link: https://rain.zeruns.com/?s=eeclub
  • RainCloud Coupon Code: zeruns
  • New users get a 50% discount for the first month. You can also claim an exclusive 20% new-user coupon in the points mall. The 20% coupon can be stacked with the official 30% annual discount, resulting in a total 56% off.

Next, select the server region and configuration based on your needs. I’m choosing Hong Kong Zone 2 with 2-core 2GB RAM. For hosting websites, we recommend selecting Hong Kong, Japan, or the US to avoid ICP备案 (mainland China registration).

Select Debian 13 as the OS, then click Purchase Now or Try—you can try it for ¥1/day.

After purchase, you can see your server under My Cloud Servers. Click Manage.

Now you’ll see your server details. Here you can reinstall/change OS or upgrade specs. Wait for the server to initialize, then proceed.

While waiting for server creation, you can prepare your large model API. Here are my recommended platforms:

I’ll use Ucloud’s AstraFlow platform for demonstration. Register via the link below and log in to the AstraFlow platform. Click API Keys at the bottom-left, then Create API Key.

Set a name for your API Key, optionally set a budget cap, leave other settings default, then click Confirm.

Now, copy and save your API Key—you’ll need it during Hermes setup. The process is similar for other platforms.

Connecting to the Server

Download and open an SSH client. Recommended clients: PuTTY or MobaXterm.

SSH client download: https://www.123pan.com/ps/2Y9Djv-UAtvH.html

I’ll use MobaXterm. In the SSH client, enter your server’s IP address (from the dashboard), SSH port (default 22), then click OK or Open.

Enter username (usually root), press Enter, then enter the password (from the dashboard, not visible while typing), press Enter.

Tip: In SSH terminal, hold left mouse button to select text—release, then click any blank area to copy. Right-click to paste.

APT Source Change (Not Needed for Overseas Servers)

Default APT package sources are overseas—use chsrc to switch to a domestic mirror.

In the SSH terminal, enter the following commands and press Enter to execute (lines starting with # are comments, don’t type them):

# Download and install chsrc
curl https://chsrc.run/posix | bash

# Auto-speed test and switch to fastest mirror
chsrc set debian

Installing and Configuring Hermes Agent

In the SSH terminal, enter this command and press Enter:

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

Wait for installation to complete.

When you see the prompt below, installation is complete and configuration has started. Press Ctrl+C to exit the setup wizard, then run hermes setup in the terminal to restart the configuration.

You’ll be asked whether to use quick setup or full setup. Choose the default—Quick Setup—by pressing Enter.

Next, select your LLM API provider. Use the up/down (↑↓) keys to select Custom endpoint (enter URL manually), then press Enter.

Enter the Ucloud API URL. From AstraFlow’s model gallery, click “API Reference” on any model to find the API URL. The mainland China API endpoint is usually https://api.modelverse.cn/v1. Enter it and press Enter.

Next, enter your API Key (the one created earlier). Input will be hidden. Press Enter after entering.

Next, select the AI model. The system will usually fetch the model list via API—just enter the model number or ID. You can also copy a model ID from the platform (e.g., deepseek-v4-flash). I’ll use this model, which offers extremely high cache hit rate and low cost—great value for AI Agent use.

Set context length—just press Enter for automatic detection.

Set a display name—press Enter to accept the default.

Proceed to message platform setup—press Enter to configure now.

Select your messaging platform—press Space to select. I’ll pick QQ Bot, then press Enter.

Choose the first option: “Scan

Missing features, such as the ability to send and receive files via a QQ bot, could be added by allowing it to modify its own code—enabling you to directly send files to it over QQ, or receive files from it. You could also add the DeepSeek-V4-Pro model for task planning and other purposes.

Some system messages are in English; you could even have it translate its own program interface into Chinese, offering high customizability and fun.

That concludes the tutorial—explore more possibilities on your own!

Recommended Reading

English Version of the Article: https://blog.zeruns.top/archives/90.html