@thomasfosterau/effect-ai-mistral

An Effect AI provider for Mistral. Effect ships no @effect/ai-mistral package, so this implements the LanguageModel service directly against Mistral's OpenAI-compatible /v1/chat/completions API.

• MistralClient — the low-level HTTP client (regular + streaming), every failure mapped to AiError. Its apiUrl override targets a proxy or any OpenAI-compatible endpoint.
• MistralLanguageModel — the LanguageModel implementation: prompt / tool / structured-output conversion and response decoding, exposed through layer, model, and make.

Install

pnpm add @thomasfosterau/effect-ai-mistral effect

effect (>=4.0.0-rc.118) is a peer dependency (it must be a singleton).

Usage

import { Effect, Layer, Redacted } from "effect";
import { FetchHttpClient } from "effect/http";
import { LanguageModel } from "effect/ai";
import { MistralClient, MistralLanguageModel } from "@thomasfosterau/effect-ai-mistral";

const ModelLayer = MistralLanguageModel.layer({ model: "mistral-large-latest" }).pipe(
  Layer.provide(MistralClient.layer({ apiKey: Redacted.make(process.env.MISTRAL_API_KEY!) })),
  Layer.provide(FetchHttpClient.layer),
);

const program = LanguageModel.generateText({ prompt: "Explain Effect in one line" }).pipe(
  Effect.map((response) => response.text),
  Effect.provide(ModelLayer),
);

Structured output uses Mistral's native json_schema response format; tool-call parameters are returned as parsed JSON and validated against the toolkit. Streaming is available via LanguageModel.streamText.

License

MIT