Integrations

Connect terminal-first engineering agents to NexoAI with one named API key and the gateway's OpenAI- or Anthropic-compatible API. Each guide keeps the provider ID nexoai, uses the public NexoAI base, and stores credentials in the harness's own configuration.

Hermes

A persistent terminal agent from Nous Research with model switching, tools, and messaging gateways.

Configure Hermes ->OpenClaw

A self-hosted agent gateway for terminal work, browser control, and optional chat channels.

Configure OpenClaw ->OpenCode

A coding agent TUI that supports custom providers through the AI SDK provider interface.

Configure OpenCode ->

One gateway, your harness

Any agent that accepts a custom OpenAI-compatible or Anthropic-compatible base URL, API key, and free-form model ID can usually use NexoAI. The core mapping is:

Harness field	NexoAI value
Provider ID	nexoai
Gateway root	https://api.nexoai.ma
OpenAI-compatible API base	https://api.nexoai.ma/v1
API key	A named dashboard key such as sk-your-nexoai-key
Model	An exact ID from the model catalog

NexoAI keys are simple named virtual keys. Create one in the dashboard, optionally add RPM, TPM, or budget caps, and configure it directly in the harness. There are no token groups and no switching application to install.

Keep the key out of repositories

Use the harness's user-level credential file or an environment variable. Never commit a live NexoAI key in a project config, shell history, screenshot, or support log.

For a client not listed here, start with the generic OpenAI-compatible setup.