pi-opengateway

pi extension that registers an OpenGateway provider for the pi coding agent. Uses the OpenAI-compatible completions API.

Install

pi install npm:@neilurk12/pi-opengateway

For project-local install:

pi install -l @neilurk12/pi-opengateway

Or from local checkout (development):

pi install /absolute/path/to/pi-opengateway

Configuration

Create a config file at one of these paths (first match wins):

1. Global: ~/.pi/agent/opengateway.json
2. Project: .pi/opengateway.json

See config.example.json for the required shape:

{
  "baseUrl": "https://your-gateway.example.com/v1",
  "apiKey": "your-api-key-here",
  "models": [
    {
      "id": "model-id",
      "name": "Model Display Name",
      "reasoning": false,
      "input": ["text"],
      "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 },
      "contextWindow": 128000,
      "maxTokens": 4096
    }
  ]
}

Fields

| Field | Required | Description | |-------|----------|-------------| | baseUrl | Yes | Gateway API endpoint (include /v1 or equivalent) | | apiKey | Yes | API key for authentication | | models[] | Yes | Array of model definitions | | models[].id | Yes | Model ID sent to the API | | models[].name | Yes | Display name shown in pi | | models[].reasoning | Yes | Whether the model supports extended thinking | | models[].input | Yes | Supported input types: ["text"] or ["text", "image"] | | models[].cost | Yes | Token costs (can be zeroes) | | models[].contextWindow | Yes | Max context window in tokens | | models[].maxTokens | Yes | Max output tokens |

Usage

Once configured, the gateway models appear in pi's model picker under the "OpenGateway" provider. Select them like any other model.

Development

npm install
npm run build

Notes

• Extensions run with full system permissions. Review code before installing any pi package.
• Config is loaded at startup. Restart pi after changing config.