Run Agents

Run Locally

Start a developer-owned Harness server with the public Hono example.

The Harness examples are a complete developer-owned Hono host: agent composition, provider adapter, local persistence, and Studio protocol implementation all remain visible in application code.

git clone https://github.com/nylorun/harness.git
cd harness
npm ci
cd examples
npm install
cp .env.example .env

Configure an OpenAI-compatible provider in .env. The endpoint, key, model, and optional headers are read only by the agent server.

NYLO_PROVIDER=OpenRouter
NYLO_MODEL=google/gemini-2.5-flash
NYLO_BASE_URL=https://openrouter.ai/api/v1
NYLO_API_KEY=your-key-here

Start the development server:

npm run dev

It listens on http://127.0.0.1:4111. The example's adapter can also target OpenAI, OpenAI-compatible gateways, or local compatible endpoints. Choose a model that supports tool calling and keep all provider credentials in the host environment, never in Studio configuration or manifests.

For a production-like process, build and start the host with npm run build then npm run start. The example README documents the complete local workflow.