Hermes Agent Quick Start | ollama launch hermes

<img src="/images/dock_head_s.png" alt="鴨編" width="24" style="vertical-align: middle;"> Quick summary: Hermes is an autonomous AI Agent by Nous Research — 70+ built-in skills, cross-session memory, and automatic skill creation. Launching it with Ollama is just one command:

brew upgrade ollama → ollama launch hermes → pick a cloud model → start chatting

---

What is Hermes Agent?

Hermes is an open-source AI Agent built by Nous Research. Unlike a plain chatbot, Hermes acts autonomously:

Feature	Description
Automatic skill creation	Builds reusable skills on the fly when it encounters new tasks
Cross-session memory	Remembers what you’ve said even after the session ends
70+ built-in skills	Search, summarize, generate code, and more — out of the box
Messaging integrations	Connect Telegram, Discord, Slack, WhatsApp, Signal, or Email

Best of all: Ollama supports ollama launch hermes natively — one command handles everything.

---

Step 1: Verify Your Ollama Version

ollama launch hermes requires a recent Ollama build. Check first:

ollama --version

If you see Error: unknown integration: hermes, or your version is below 0.7.0, update:

brew upgrade ollama

After updating, restart Ollama (click the menu-bar icon → Quit, then reopen it).

---

Step 2: Launch Hermes in One Command

Open Terminal and run:

ollama launch hermes

On the first run, Ollama asks if you want to install Hermes — press Enter to confirm:

Ollama then automatically runs four steps:

1. Install — installs Hermes Agent via npm
2. Model — opens the model selector
3. Onboarding — configures Ollama as the provider, pointing Hermes at http://127.0.0.1:11434/v1
4. Launch — starts the Hermes chat interface

---

Step 3: Choose a Model

After installation, Hermes shows you a model selector.

<img src="/images/dock_head_s.png" alt="鴨編" width="24" style="vertical-align: middle;"> Recommended cloud models (uses Ollama’s free quota — no local download needed):

Model	Strengths
minimax-m2.7:cloud	🥇 Fast and efficient, great for everyday productivity
kimi-k2.5:cloud	Strong reasoning, supports subagents
glm-5.1:cloud	Reasoning and code generation
qwen3.5:cloud	Tool use + vision, great for agentic tasks

Prefer to run fully offline? Local models require a GPU:

Model	Requirement
gemma4	~16 GB VRAM
qwen3.6	~24 GB VRAM

Select your model and Hermes finishes setup automatically.

---

Step 4: Your First Conversation

Once Hermes is running, type anything you’d like it to do:

Hi! Please introduce yourself — who is Hermes, and what can you do?

The key difference from a regular chatbot: Hermes saves a summary of your conversation to memory, so the next time you start it up, it still knows what you talked about.

---

Step 5 (Optional): Connect a Messaging Platform

Want to chat with Hermes from Telegram or Discord? Run:

hermes gateway setup

Supported platforms: Telegram, Discord, Slack, WhatsApp, Signal, Email.

Once set up, you can message Hermes directly from your preferred app — no need to open a terminal.

---

Troubleshooting

🚨 Error: unknown integration: hermes

Your Ollama version is too old. Run brew upgrade ollama to update, then try again.

🚨 Stuck at the npm install step

Hermes requires Node.js 18 or higher. Check:

node --version

If missing:

brew install node

🚨 Want to reconfigure the model or provider

Re-run the full setup wizard at any time:

hermes setup

---

Appendix: Manual Installation

Prefer to drive the setup yourself instead of using ollama launch? Install Hermes directly:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

The setup wizard launches automatically. To connect to Ollama:

1. Select More providers…

2. Select Custom endpoint (enter URL manually)

3. Enter the API base URL:

   http://127.0.0.1:11434/v1

4. Leave the API key blank (not required for local Ollama)

5. Hermes auto-detects your downloaded models — confirm and press Enter

Windows users: Hermes on Windows requires WSL2. Install it with wsl --install and run everything from inside the WSL shell. See the Windows guide.

---

What’s Next

• 🪟 Hermes on Windows (WSL2)
• 🦀 Meet OpenClaw: another AI Agent framework
• 📱 Connect OpenClaw to Telegram

Questions? Join the discussion at the homepage!