--- name: mistral-sdk-patterns-6de74ba49f6d description: | Apply production-ready Mistral AI SDK patterns for TypeScript and Python. Use when implementing Mistral integrations, refactoring SDK usage, or establishing team coding standards for Mistral AI. Trigger with phrases like "mistral SDK patterns", "mistral best practices", "mistral code patterns", "idiomatic mistral". allowed-tools: Read, Write, Edit version: 1.0.0 license: MIT author: Jeremy Longshore --- # Mistral AI SDK Patterns ## Overview Production-ready patterns for Mistral AI SDK usage in TypeScript and Python. ## Prerequisites - Completed `mistral-install-auth` setup - Familiarity with async/await patterns - Understanding of error handling best practices ## Instructions ### Step 1: Implement Singleton Pattern (Recommended) **TypeScript** ```typescript // src/mistral/client.ts import Mistral from '@mistralai/mistralai'; let instance: Mistral | null = null; export interface MistralConfig { apiKey: string; timeout?: number; maxRetries?: number; } export function getMistralClient(config?: Partial): Mistral { if (!instance) { const apiKey = config?.apiKey || process.env.MISTRAL_API_KEY; if (!apiKey) { throw new Error('MISTRAL_API_KEY is required'); } instance = new Mistral({ apiKey, timeout: config?.timeout ?? 30000, }); } return instance; } // For testing - reset singleton export function resetMistralClient(): void { instance = null; } ``` **Python** ```python # src/mistral/client.py import os from mistralai import Mistral from typing import Optional _instance: Optional[Mistral] = None def get_mistral_client(api_key: Optional[str] = None) -> Mistral: global _instance if _instance is None: key = api_key or os.environ.get("MISTRAL_API_KEY") if not key: raise ValueError("MISTRAL_API_KEY is required") _instance = Mistral(api_key=key) return _instance def reset_mistral_client() -> None: global _instance _instance = None ``` ### Step 2: Add Error Handling Wrapper **TypeScript** ```typescript import Mistral from '@mistralai/mistralai'; interface MistralResult { data: T | null; error: Error | null; usage?: { promptTokens: number; completionTokens: number; totalTokens: number }; } async function safeMistralCall( operation: () => Promise ): Promise> { try { const data = await operation(); return { data, error: null }; } catch (err) { const error = err as Error; console.error({ name: error.name, message: error.message, // Extract status code if available status: (err as any).status, }); return { data: null, error }; } } // Usage const result = await safeMistralCall(() => client.chat.complete({ model: 'mistral-small-latest', messages: [{ role: 'user', content: 'Hello!' }], }) ); if (result.error) { console.error('Chat failed:', result.error.message); } else { console.log(result.data?.choices?.[0]?.message?.content); } ``` ### Step 3: Implement Retry Logic with Exponential Backoff ```typescript interface RetryConfig { maxRetries: number; baseDelayMs: number; maxDelayMs: number; jitterMs: number; } async function withRetry( operation: () => Promise, config: RetryConfig = { maxRetries: 3, baseDelayMs: 1000, maxDelayMs: 32000, jitterMs: 500 } ): Promise { for (let attempt = 0; attempt <= config.maxRetries; attempt++) { try { return await operation(); } catch (err) { if (attempt === config.maxRetries) throw err; const status = (err as any).status; // Only retry on rate limits (429) or server errors (5xx) if (status !== 429 && (status < 500 || status >= 600)) throw err; const exponentialDelay = config.baseDelayMs * Math.pow(2, attempt); const jitter = Math.random() * config.jitterMs; const delay = Math.min(exponentialDelay + jitter, config.maxDelayMs); console.log(`Attempt ${attempt + 1} failed. Retrying in ${delay.toFixed(0)}ms...`); await new Promise(r => setTimeout(r, delay)); } } throw new Error('Unreachable'); } ``` ### Step 4: Type-Safe Chat Wrapper ```typescript import Mistral from '@mistralai/mistralai'; type MistralModel = | 'mistral-large-latest' | 'mistral-medium-latest' | 'mistral-small-latest' | 'open-mistral-7b' | 'open-mixtral-8x7b'; interface ChatOptions { model?: MistralModel; temperature?: number; maxTokens?: number; topP?: number; stream?: boolean; } interface Message { role: 'system' | 'user' | 'assistant'; content: string; } export async function chat( client: Mistral, messages: Message[], options: ChatOptions = {} ): Promise { const response = await client.chat.complete({ model: options.model ?? 'mistral-small-latest', messages, temperature: options.temperature, maxTokens: options.maxTokens, topP: options.topP, }); return response.choices?.[0]?.message?.content ?? ''; } ``` ### Step 5: Streaming Helper ```typescript export async function* streamChat( client: Mistral, messages: Message[], options: ChatOptions = {} ): AsyncGenerator { const stream = await client.chat.stream({ model: options.model ?? 'mistral-small-latest', messages, temperature: options.temperature, maxTokens: options.maxTokens, }); for await (const event of stream) { const content = event.data?.choices?.[0]?.delta?.content; if (content) { yield content; } } } // Usage for await (const chunk of streamChat(client, messages)) { process.stdout.write(chunk); } ``` ## Output - Type-safe client singleton - Robust error handling with structured logging - Automatic retry with exponential backoff - Streaming support with async generators ## Error Handling | Pattern | Use Case | Benefit | |---------|----------|---------| | Safe wrapper | All API calls | Prevents uncaught exceptions | | Retry logic | Transient failures | Improves reliability | | Type guards | Response validation | Catches API changes | | Logging | All operations | Debugging and monitoring | ## Examples ### Factory Pattern (Multi-tenant) ```typescript const clients = new Map(); export function getClientForTenant(tenantId: string): Mistral { if (!clients.has(tenantId)) { const apiKey = getTenantApiKey(tenantId); clients.set(tenantId, new Mistral({ apiKey })); } return clients.get(tenantId)!; } ``` ### Python Context Manager ```python from contextlib import contextmanager from mistralai import Mistral @contextmanager def mistral_client(): client = Mistral(api_key=os.environ.get("MISTRAL_API_KEY")) try: yield client finally: pass # Cleanup if needed # Usage with mistral_client() as client: response = client.chat.complete( model="mistral-small-latest", messages=[{"role": "user", "content": "Hello!"}] ) ``` ### Zod Response Validation ```typescript import { z } from 'zod'; const mistralResponseSchema = z.object({ id: z.string(), object: z.literal('chat.completion'), model: z.string(), choices: z.array(z.object({ index: z.number(), message: z.object({ role: z.enum(['assistant']), content: z.string(), }), finishReason: z.string(), })), usage: z.object({ promptTokens: z.number(), completionTokens: z.number(), totalTokens: z.number(), }), }); function validateResponse(response: unknown) { return mistralResponseSchema.parse(response); } ``` ## Resources - [Mistral AI API Reference](https://docs.mistral.ai/api/) - [Mistral AI Client Libraries](https://docs.mistral.ai/getting-started/clients/) - [Zod Documentation](https://zod.dev/) ## Next Steps Apply patterns in `mistral-core-workflow-a` for chat completions. --- Source attribution: Dicklesworthstone Original project: https://github.com/Dicklesworthstone/pi_agent_rust/blob/f28fc46ee0db8e710d95b8b55592cd38d60f2b26/tests/ext_conformance/artifacts/plugins-community/plugins/saas-packs/mistral-pack/skills/mistral-sdk-patterns/SKILL.md License: MIT License information: https://opensource.org/license/mit Original upstream instructions; downloaded from Mega AI Hub. ## Upstream license notice MIT License (with OpenAI/Anthropic Rider) Copyright (c) 2026 Jeffrey Emanuel Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: ADDITIONAL RIDER / RESTRICTION (OpenAI / Anthropic): This rider is part of the "conditions" of this License. In the event of any conflict between this rider and any other portion of this License, this rider controls. "Restricted Parties" means OpenAI, L.L.C.; Anthropic, PBC; any of their respective Affiliates; and any person or entity acting directly or indirectly on behalf of, for the benefit of, or under the direction of any of the foregoing (including any officer, director, employee, contractor, agent, consultant, service provider, or representative). Notwithstanding any other provision of this License, no rights are granted to any Restricted Party. Any purported license, sublicense, assignment, transfer, or other permission to any Restricted Party is null and void absent the express prior written permission of Jeffrey Emanuel. You may not provide, disclose, distribute, sublicense, sell, lease, lend, host, make available, or otherwise permit access to the Software or any derivative work of the Software (as defined in applicable copyright law) (collectively, "Derivative Works") to or for any Restricted Party. For purposes of this rider, "use" includes, without limitation: copying, modifying, merging, publishing, distributing, sublicensing, selling, transferring, making available, hosting, deploying, executing, benchmarking, testing, analyzing, indexing, or incorporating the Software or any Derivative Works into any dataset, training corpus, evaluation harness, or pipeline for machine learning or other automated systems. This rider applies to the Software and all Derivative Works. As a condition of use, you agree that this rider is a precondition to exercising any rights under this License, and you agree that any distribution of the Software or any Derivative Works must include this rider provision unmodified. Any breach of this rider automatically and immediately terminates the permissions granted by this License. Upon termination, you must immediately cease all use and distribution of the Software and any Derivative Works and destroy all copies under your control. You agree that a breach of this rider would cause irreparable harm and that Jeffrey Emanuel may seek injunctive or other equitable relief to enforce this rider, in addition to any other remedies available at law. To the maximum extent permitted by applicable law, the prevailing party in any action to enforce this rider shall be entitled to recover reasonable attorneys' fees and costs. For purposes of this rider, "Affiliate" means any entity that directly or indirectly controls, is controlled by, or is under common control with the specified party. "Control" means ownership of more than 50% of the voting securities or other ownership interest, or the power to direct management or policies by contract or otherwise. The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. # Mega AI Hub resource metadata Creator / publisher: Dicklesworthstone Original project: https://github.com/Dicklesworthstone/pi_agent_rust/blob/f28fc46ee0db8e710d95b8b55592cd38d60f2b26/tests/ext_conformance/artifacts/plugins-community/plugins/saas-packs/mistral-pack/skills/mistral-sdk-patterns/SKILL.md Attribution basis: Source repository record Fields of work: Development & Engineering Roles: Backend Developer, Data Engineer, ML / AI Engineer Specialties: API, Backend Tools: Mistral Country applicability: Country not specified Classification basis: Source catalog labels Instruction SHA-256: 98626d028f76d82cfc8edcefa9be5179672f05d0e552a689e91f2cbcbbb93ee8