mistral

Mistral AI API Skill

Overview

Mistral AI provides frontier open and commercial AI models, featuring mistral-large-latest, mistral-small-latest, codestral-latest for coding, and mistral-embed for semantic search.

Installation

npm install @mistralai/mistralai
pip install mistralai

Authentication & Client Setup

import { Mistral } from '@mistralai/mistralai';

const client = new Mistral({
  apiKey: process.env.MISTRAL_API_KEY,
});

Core API Operations

1. Chat Completion with Tool Calling

const response = await client.chat.complete({
  model: 'mistral-large-latest',
  messages: [
    { role: 'user', content: 'What is the stock price of AAPL?' },
  ],
  tools: [
    {
      type: 'function',
      function: {
        name: 'getStockPrice',
        description: 'Get the current stock price for a ticker',
        parameters: {
          type: 'object',
          properties: {
            ticker: { type: 'string' },
          },
          required: ['ticker'],
        },
      },
    },
  ],
  toolChoice: 'auto',
});

2. Code Generation (Codestral)

const codeCompletion = await client.chat.complete({
  model: 'codestral-latest',
  messages: [
    { role: 'system', content: 'You are an expert TypeScript developer.' },
    { role: 'user', content: 'Write a debounce function in TypeScript with type generics.' },
  ],
});

3. Generate Semantic Text Embeddings

const embeddings = await client.embeddings.create({
  model: 'mistral-embed',
  inputs: ['Document chunk 1', 'Document chunk 2'],
});

console.log(embeddings.data[0].embedding.length); // 1024

AI Pitfalls & Anti-Hallucination Guidelines

• Codestral Endpoint Confusion: Codestral has dedicated API keys and rate limits when accessed via codestral.mistral.ai.
• JSON Mode Requirement: When requesting JSON mode, the word "json" must explicitly appear in the prompt instructions.

Production Checklist

• Structured JSON responses verified with Zod
• Safe fallback for tool calls without infinite agent loops
• Rate limits handled on embedding batch calls