Skip to content
KoboiLLM Docs

Hermes Agent Integration

Hermes Agent is an autonomous AI coding agent developed by Nous Research. It operates directly in your CLI or background environment with capabilities including terminal execution, file system manipulation, architecture diagram rendering, and background computer use.

Connecting Hermes Agent to KoboiLLM enables autonomous agentic workflows using high-performance models (such as GPT-4o, Claude 3.5 Sonnet, and Gemini 2.5 Flash) with prompt caching and strict budget controls.

Step 1: Obtain Your KoboiLLM API Key

  1. In your KoboiLLM dashboard, navigate to Virtual Keys → + Create New Key.
  2. Label your key (e.g., hermes_agent_key), select All Team Models, choose Default, and click Create Key.
  3. Copy the secret key string (sk-xxxxxx).

Step 2: Configuration Methods

Method A: Interactive CLI Wizard

  1. Open your terminal and run:
    Terminal window
    hermes model
  2. Select Custom Endpoint from the prompt menu.
  3. Supply the following inputs:
    • Base URL: https://api.koboillm.com/v1
    • API Key: Your KoboiLLM Virtual Key (sk-xxxxxx).
    • Model Name: Desired default model (e.g., openai/gpt-4o).
    • Supports Vision: Enter y.

Method B: Manual Configuration via config.yaml

Add KoboiLLM as a custom provider in your global Hermes Agent configuration file:

  • Linux / macOS: ~/.hermes/config.yaml
  • Windows: %USERPROFILE%\.hermes\config.yaml
~/.hermes/config.yaml
custom_providers:
  - name: koboillm
    base_url: https://api.koboillm.com/v1
    key_env: KOBOILLM_API_KEY
    api_mode: chat_completions


model:
  default: openai/gpt-4o
  provider: custom:koboillm

Set the key in your shell environment:

Terminal window
export KOBOILLM_API_KEY="sk-x...
## Step 3: Run Hermes Agent


Launch Hermes Agent normally in your terminal:


```bash
hermes