详细分析 ▾
运行时依赖
安装命令
点击复制技能文档
Mistral MCP for OpenClaw
创建d by the mAIntAIner of mistral-mcp. This is a community 技能, not an official OpenClaw or Mistral integration.
Use this 技能 when you want OpenClaw to 访问 Mistral capabilities beyond the built-in chat/模型 routing 提供者:
OCR for documents and images Codestral fill-in-the-middle (FIM) code completion Voxtral transcription and speech 工具s Moderation and classification 端点s Files and batch API 工作流s Live 模型 and voice resources
OpenClaw already includes a built-in Mistral 提供者 for chat. This 技能 is for 工具-level MCP 访问 alongside that 提供者.
Requirements Node.js 18+ OpenClaw 命令行工具 MISTRAL_API_KEY in your 环境 mistral-mcp 安装ed from npm 设置up
安装 the MCP server package globally:
npm 安装 -g mistral-mcp
设置 your Mistral API key in your shell 环境:
导出 MISTRAL_API_KEY="sk-..."
Register the stdio MCP server in OpenClaw:
OpenClaw mcp 设置 mistral '{"command":"mistral-mcp","env":{"MISTRAL_API_KEY":"${MISTRAL_API_KEY}"}}'
检查 the saved definition:
OpenClaw mcp show mistral --json
When to use it
Use this 技能 for 工作流s where the 代理 needs a Mistral-specific 工具, not just a chat 模型:
提取 text from a PDF or image with OCR Ask Codestral for FIM / inline code completion Transcribe or 生成 audio with Voxtral 运行 moderation or classification before taking an action Submit larger a同步 workloads through the batch API Inspect live 模型 and voice cata记录s as MCP resources Safety notes Do not paste API keys into chat or commit them to source files. Prefer 环境 variables or your normal secret 管理器. Review the mistral-mcp package before 安装ing it if you operate in a sensitive workspace. Source: https://github.com/Swih/mistral-mcp. Mistral has its own pricing and rate limits. 检查 the current Mistral plan and usage policies before 运行ning batch or large transcription workloads.