Search docs...

Start typing to search documentation

Hermes Agent
Hermes Agent

Hermes Agent Integration

Use UnoRouter as the model provider for Hermes Agent. Point one OpenAI-compatible endpoint at UnoRouter and reach every model.

Overview

Hermes Agent by Nous Research is a self-improving AI agent that runs as a CLI, TUI, messaging gateway, or IDE integration with durable on-disk memory. Set its provider to custom and point the base URL at UnoRouter to route every turn through one OpenAI-compatible endpoint that serves all models.

Quick Config

Drop these values into the client. Your key auto-fills when you are signed in.

text
Base URL: https://api.unorouter.com/v1
API Key: YOUR_API_KEY

Sign in to auto-fill your API key

Compatibility

Chat CompletionsStreamingTool calling

Step-by-step setup

  1. 1
    Install Hermes

    Run the install script to set up Hermes Agent. On Windows, install it inside WSL2.

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

    Sign in to auto-fill your API key

  2. 2
    Point the provider at UnoRouter

    Edit `~/.hermes/config.yaml`, set `model.provider` to `custom`, and set `model.base_url` to UnoRouter's base URL. Pick a default model with at least 64k context.

    yaml
    # ~/.hermes/config.yaml
    model:
      provider: "custom"
      base_url: "https://api.unorouter.com/v1"
      default: "YOUR_MODEL_ID"

    Sign in to auto-fill your API key

  3. 3
    Add your API key

    Store your UnoRouter API key as `OPENAI_API_KEY` in `~/.hermes/.env` so Hermes authenticates every turn.

    bash
    # ~/.hermes/.env
    OPENAI_API_KEY=YOUR_API_KEY

    Sign in to auto-fill your API key

  4. 4
    Run Hermes

    Start the TUI with `hermes --tui` and begin working.

powershell
# Hermes runs under WSL2 on Windows.
# In ~/.hermes/config.yaml set model.provider: "custom"
# and model.base_url: "https://api.unorouter.com/v1"
export OPENAI_API_KEY="YOUR_API_KEY"

hermes --tui

Sign in to auto-fill your API key

Recommended models

Free models that work well here. Paste any model id into the client.

lfm-2.5-1.2b-instructqwen2.5-coder-32bmistral-small-2603nemotron-3.5-content-safetygemini-pro-latest

Gotchas

  • Hermes rejects models under 64,000 tokens of context at startup. Pick a larger-context model from UnoRouter's models page.
  • If startup fails its `/v1/models` check, confirm `model.base_url` ends in `/v1` and `OPENAI_API_KEY` is set in `~/.hermes/.env`.
Generate API KeyModels