mistral-core-workflow-a

# Mistral AI Core Workflow A: Chat Completions ## Overview Production chat completion patterns for Mistral AI: multi-turn conversations, streaming responses, JSON mode structured output, guardrails/moderation, and model selection. Uses the `@mistralai/mistralai` SDK. ## Prerequisites - Completed `mistral-install-auth` setup - `MISTRAL_API_KEY` environment variable set - Understanding of Mistral model tiers ## Instructions ### Step 1: Basic Chat Completion ```typescript import { Mistral } from '@mistralai/mistralai'; const client = new Mistral({ apiKey: process.env.MISTRAL_API_KEY }); async function chat(userMessage: string): Promise<string> { const response = await client.chat.complete({ model: 'mistral-small-latest', messages: [ { role: 'system', content: 'You are a helpful assistant.' }, { role: 'user', content: userMessage }, ], }); return response.choices?.[0]?.message?.content ?? ''; } ``` ### Step 2: Multi-Turn Conversation Manager ```typescript interface Message { role: 'system' | 'user' | 'assistant'; content: string; } class MistralConversation { private messages: Message[] = []; private client: Mistral; private model: string; constructor(systemPrompt: string, model = 'mistral-small-latest') { this.client = new Mistral({ apiKey: process.env.MISTRAL_API_KEY }); this.model = model; this.messages.push({ role: 'system', content: systemPrompt }); } async send(userMessage: string): Promise<string> { ...