Monorch

@monorch/ai/openai

OpenAI-compatible chat completions provider and a scripted mock for tests. Also re-exported from @monorch/ai.

Import

import.ts
import { openai, mock } from "@monorch/ai/openai";
// or: import { openai, mock } from "@monorch/ai";

Constructors

openai(modelId, options?)
Returns a ModelProvider with generate and SSE stream. Point baseUrl at LiteLLM, OpenRouter, Azure-compatible, or local gateways.
mock(script?)
Scripted turns for smoke and unit tests. Each generate consumes the next GenerateResult. Respects abort.

OpenAiOptions

apiKey?
Bearer token. Falls back to env conventions in examples.
baseUrl?
API root ending in /v1 style chat completions.
timeoutMs?
Default 60_000. Merged with AbortSignal on fetch.
defaultHeaders?
Extra headers on every request.

Keywords

Provider stream chunks: text, tool_call, done. Agent loops prefer provider stream() when present and emit progressive text AiEvents.

Guide: Providers.

FAQ