AI Agents and Automation
The Hermes Agent Setup Guide
The fastest project in GitHub history to hit 100,000 stars. Built by Nous Research. Install in 5 minutes and run a full AI agent loop from your desktop.
Hermes is an open-source AI agent framework from Nous Research, the fastest GitHub project to hit 100,000 stars. It runs a full autonomous loop from your desktop: plan a task, execute steps, observe results, adjust, repeat. No cloud subscription required. Setup takes about 5 minutes and the first successful agent run usually follows within 10.
How to do it
- 1Clone the Hermes repo from GitHub: `git clone https://github.com/NousResearch/hermes-agent` (requires Python 3.10 or later)
- 2Install dependencies: `pip install -r requirements.txt` inside the project folder
- 3Add your API key to `.env`. OpenAI, Anthropic, and local Ollama models are all supported
- 4Run `python main.py` and describe a task in plain English. Start with something simple like "research the top 5 AI frameworks and summarise each in 3 bullet points"
- 5Watch the loop: the agent plans, executes each step, observes what happened, and adjusts until the task is done
- 6Configure tool access in `config.yaml`. Toggle web search, code execution, and file system permissions based on what you need
Why it works
Most AI tools are single-turn: you prompt, it answers, done. Hermes runs a loop. It plans, acts, sees what happened, and adjusts until the task is finished. The 100,000-star milestone happened quickly because it actually works on real tasks out of the box without needing a paid subscription or cloud infrastructure.
All guides are free, no sign-up required. Browse free resources