@moxxy/plugin-provider-openai

The OpenAI LLMProvider for moxxy — streams Chat Completions with tool_calls and translates moxxy's ProviderRequest/event shape ↔ the OpenAI wire format. Register it into a Session to route turns to OpenAI.

Because it speaks the OpenAI Chat Completions API, it also drives any OpenAI-compatible endpoint (z.ai, xAI, Google's OpenAI shim, Ollama / local) via the baseURL override.

Install

npm i @moxxy/core @moxxy/sdk @moxxy/mode-default @moxxy/plugin-provider-openai

Usage

import { Session, collectTurn, autoAllowResolver } from '@moxxy/core';
import defaultModePlugin from '@moxxy/mode-default';
import openaiPlugin from '@moxxy/plugin-provider-openai';

const session = new Session({ cwd: process.cwd(), permissionResolver: autoAllowResolver });
session.pluginHost.registerStatic(defaultModePlugin);
session.pluginHost.registerStatic(openaiPlugin);

session.providers.setActive('openai', {
  apiKey: process.env.OPENAI_API_KEY,
  // model: 'gpt-5.2',          // optional — defaults to the built-in catalog
});

console.log((await collectTurn(session, 'Hello!')).findLast((e) => e.type === 'assistant_message')?.content);

Point at an OpenAI-compatible vendor

session.providers.setActive('openai', {
  apiKey: process.env.GROQ_API_KEY,
  baseURL: 'https://api.groq.com/openai/v1',
  // model: '...'   // the vendor's model id
});

Config (OpenAIProviderConfig)

| field | meaning | |---|---| | apiKey | the API key (or wire it through the Session's secretResolver) | | baseURL | override the API base for an OpenAI-compatible vendor | | model / models | pick / override the advertised model catalog |

Exports

default / openaiPlugin (register this), openaiProviderDef, OpenAIProvider, openAIModels, validateKey, and defineOpenAICompatProvider (build a provider for another OpenAI-compatible vendor in a few lines).

License

MIT