Connect Hermes Agent to OpenLLM Buddy

This guide walks through connecting Hermes Agent to your OpenAI-compatible OpenLLM Buddy endpoint so it can run gemma4:26b as the inference backend.

Integration guide

What is Hermes Agent?

Hermes Agent (by Nous Research) is an open-source autonomous agent that runs on your machine, remembers context over time, and connects to chat platforms like Telegram, Discord, and Slack. It is not tied to a single cloud API — you choose the inference provider.

Why connect OpenLLM Buddy?

OpenLLM Buddy exposes an OpenAI-compatible API. Hermes supports a Custom OpenAI-compatible endpoint during setup, so you can point it at your deployment and use your own API keys and models.

How this guide is structured

  • Install — official install script
  • Setup wizard — hermes setup
  • API details — base URL, key, and model name
  • Optional — config.yaml or environment variables
  • Launch — run hermes

Create your API key in the console after you deploy a model instance.

1

Install Hermes Agent

Install Hermes Agent using the official script from hermes-agent.nousresearch.com. Run this in your terminal (macOS or Linux).

Terminal
bash
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

When the install finishes, continue with hermes setup in the next step.

2

Run setup

Start the interactive setup wizard. When Hermes prompts you for an inference provider, select Custom OpenAI-compatible endpoint and follow the prompts.

Terminal
bash
hermes setup
  1. Choose Custom OpenAI-compatible endpoint (not a built-in cloud provider).
  2. Hermes will walk you through base URL, API key, and model name in the next step.

You can also skip the wizard and configure Hermes manually via config.yaml or environment variables — see the optional steps below.

3

Enter your API details

During hermes setup, provide the following when prompted. Hermes verifies the endpoint against /v1/models and confirms the detected model before continuing.

Base URL
https://openllmbuddy-proxy.botbuddytech.workers.dev/v1

Use the /v1 root — not /v1/chat/completions.

API key
YOUR_API_KEY

Create one in the console (starts with ob_sk_).

Model name
gemma4:26b

Examples: Gemma 4 26B → gemma4:26b, Qwen 3.6 27B → qwen3.6:27b

If verification fails, double-check that the base URL uses https:// and ends with /v1, and that your API key is active for a running deployment.

4

Configure via config.yaml (optional)

If you prefer editing the config file directly instead of the setup wizard, add a model block like this:

config.yaml
yaml
model:
  provider: custom
  base_url: https://openllmbuddy-proxy.botbuddytech.workers.dev/v1
  api_key: YOUR_API_KEY
  model: gemma4:26b

When base_url is set, Hermes calls that endpoint directly and uses api_key or the OPENAI_API_KEY environment variable for authentication.

5

Use environment variables (optional)

Alternatively, export standard OpenAI-compatible environment variables before launching Hermes:

Shell
bash
export OPENAI_BASE_URL=https://openllmbuddy-proxy.botbuddytech.workers.dev/v1
export OPENAI_API_KEY=YOUR_API_KEY

Replace the placeholder API key with your real key from the console. These variables work well with Docker, systemd, or launchd setups.

6

Launch Hermes

Once setup is complete (wizard, config.yaml, or env vars), start the agent:

Terminal
bash
hermes

Hermes will use your custom endpoint for inference. You can then connect chat channels, skills, and scheduled tasks from the Hermes docs on hermes-agent.nousresearch.com.