Connect Hermes to LiteLLM with a virtual key

Hermes Agent can use LiteLLM as a named custom provider. This setup sends model requests to the gateway's /v1/chat/completions endpoint with your LiteLLM virtual key.

Configure the connection

Create a virtual key with access to a model that supports tool calls. Add the following to ~/.hermes/config.yaml, merging it with your existing settings:

~/.hermes/config.yaml

providers:
  litellm:
    api: https://litellm.example.com/v1
    transport: chat_completions
    key_env: LITELLM_API_KEY

model:
  default: my-model
  provider: custom:litellm

Replace the gateway URL and my-model with a model_name available to your key. The URL includes /v1; Hermes appends /chat/completions.

Export the key in the terminal where you launch Hermes:

export LITELLM_API_KEY="sk-<your-virtual-key>"
hermes

For a persistent configuration, put LITELLM_API_KEY in ~/.hermes/.env, which Hermes loads for its credentials. Keep the key out of config.yaml and source control.

Verify

Start a new Hermes session and send a prompt. Confirm the request in LiteLLM's Admin UI under Logs or Usage, attributed to your virtual key and the selected model. Test a tool call with your chosen model before using it for agent tasks.

For sign-in through your organization's identity provider, use Hermes SSO. To add conversation tracing, see the separate Hermes Lens integration.