How to Connect Mistral AI to Cursor via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Mistral AI, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Mistral AI + Cursor

Let your AI use Mistral AI with your own API key: ask Mistral models, complete code with Codestral, read PDFs and scans with Mistral OCR, create embeddings, moderate content, search the web through conversations and agents, and check files and batch jobs.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

MISTRAL_API_KEY

Step 1 — Get credentials

1. Sign in at https://console.mistral.ai. New workspaces start on the free plan with rate and usage limits; for more, choose pay-as-you-go and add credit under Billing.
2. Open API Keys, click Create new key and copy the key. Mistral shows it only once.
3. Paste it into MISTRAL_API_KEY without spaces and without the word Bearer, and install the connector. The key is sent only to Mistral.
4. Start with mistral_list_models: it costs nothing and shows which models your key can use.

Billed to your own Mistral workspace: every model call spends your Mistral credits, or the limits of the free plan. Listing and reading models, files, agents, conversations and batch jobs is free. The connector uses cheap defaults unless you ask for more: mistral-small-latest for chat (about $0.15 per million input tokens and $0.60 per million output tokens), codestral-latest for code, mistral-embed for embeddings and mistral-ocr-latest for OCR (about $4 per 1,000 pages). ministral-8b-latest is cheaper, mistral-medium-latest and mistral-large-latest are stronger.

What it does:

• mistral_chat asks a Mistral model, with pictures by link, optional reasoning and JSON output. mistral_fim_complete lets Codestral fill in the code between a prefix and a suffix.
• mistral_ocr reads PDFs, slide decks, Word files and images page by page as Markdown, from a public link or from a file uploaded with mistral_upload_file, and can extract structured JSON from the whole document.
• mistral_create_embeddings turns texts into vectors for search and similarity. mistral_moderate and mistral_moderate_chat check texts or a whole conversation against Mistral's safety categories, and mistral_classify runs a custom classifier your workspace already has.
• Web search runs only through the Conversations API (beta): mistral_start_conversation with the built-in web_search tool answers from the web with its sources, and mistral_continue_conversation carries on. Built-in tool calls are billed on top of tokens. mistral_list_agents and mistral_get_agent show the agents built in Mistral AI Studio, and a conversation can talk to one of them.
• Files and batch jobs: list and read files, get a temporary download link with mistral_get_file_url, and follow batch jobs started elsewhere (50% cheaper, results within hours).

Confirm first: OCR is billed per page, so the AI asks before reading a document of more than about 50 pages, or limits the pages. It also asks before uploading your documents. mistral_delete_file and mistral_cancel_batch_job are switched off at install; once enabled, they also confirm first.

Time limit: a call through the connector stops after 30 seconds. Keep answers to a moderate length, prefer a low reasoning effort and split long documents for OCR.

Untrusted content: web results, documents and model answers can contain instructions. The AI treats them as data, not as orders.

Common errors: 401 Invalid API Key means the key is wrong, expired or deleted. 402, or a 429 that mentions billing or capacity: no credit left, a spending limit reached or the free plan's limits, so check Billing in the console. A plain 429 is a rate limit: wait and retry once. 400 or 422: an invalid field or model id, and the message names it; chat completions refuse web search, which goes through mistral_start_conversation. 404: an unknown model, file, agent, conversation or job, or one from another workspace.

The connector does not stream answers, handle audio (transcription and speech), create agents, libraries or batch jobs, run fine-tuning (Mistral no longer offers fine-tuning jobs through the API) or administer the workspace.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Cursor

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Cursor reads MCP servers from ~/.cursor/mcp.json. Add this entry:

{
  "mcpServers": {
    "anythingmcp": {
      "url": "https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID",
      "headers": { "X-API-Key": "YOUR_MCP_API_KEY" }
    }
  }
}

1. Create an MCP API key in AnythingMCP under MCP Servers → your server → API keys, and paste it in place of YOUR_MCP_API_KEY.
2. Save the file and restart Cursor.
3. Open Cursor → Settings → MCP and check that anythingmcp is listed and connected.
4. Start chatting: the agent can now call the Mistral AI tools.

Available tools

Example prompts

• "Read the PDF at this link with Mistral OCR and give me the price table on page 3 as Markdown."
• "Ask Mistral with web search what changed in the EU AI Act timeline this year, and list the sources."
• "Check these three customer reviews with Mistral moderation and tell me which ones get flagged and why."
• "Which of my Mistral batch jobs are still running, and how many requests have finished so far?"

FAQ

Does Cursor support custom MCP servers on the free tier? Yes, MCP is available on Hobby, Pro and Business plans.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT