@lightweight-ai/opencode-plugin

The @lightweight-ai/opencode-plugin connects the OpenCode CLI to the Lightweight AI API gateway. It provides seamless access to a wide range of large language models through a single unified interface.

Description

This plugin integrates the Lightweight AI API (api.lightweight.one) into OpenCode, allowing users to leverage over 50 models including GPT-5, Claude Opus 4.6, and Gemini. It handles model discovery, capability mapping, and authentication automatically.

Installation

Install the package via npm:

npm install @lightweight-ai/opencode-plugin

Getting an API Key

To use this plugin, you must register at the Lightweight AI platform to obtain an API key. Your API key should follow the format lw_sk_....

Configuration

The plugin requires the LIGHTWEIGHT_API_KEY environment variable to be set.

export LIGHTWEIGHT_API_KEY=lw_sk_your_api_key_here

Usage

Once installed and configured, OpenCode will automatically detect the "lightweight" provider. The plugin fetches available models from the API and caches them for one hour.

Reasoning models automatically include effort level variants:

• low
• medium
• high
• xhigh

How it Works

The plugin implements two primary hooks:

• config: Registers the "lightweight" provider using @ai-sdk/openai as the underlying SDK. It points to https://api.lightweight.one/v1 and maps models based on their reported capabilities (tool calling, attachments, reasoning).
• chat.headers: Adds X-Client: opencode and X-Plugin-Version: 1.0.0 headers to all outgoing requests for better telemetry and support.

Model discovery is performed by fetching /v1/models with the provided Bearer token. The plugin maps technical constraints such as context and output limits automatically.

License

MIT