Mistral AI for AI agents

Mistral AI is a research lab providing open-weight and hosted large language models through an API for chat, embeddings, and OCR. Connect it once through Arc0, and your agent, or Claude, ChatGPT and Cursor, can use it through one MCP endpoint, limited to what each user approved.

AI and modelsAPI keyMCP + RESTmistral.ai
AUDIT LOG · MISTRAL AIPOLICY: acme-support
09:41:07 · claude · u_8f2read
mistral_ai.list_files
List Files✓ allowed · 212ms
09:41:08 · claude · u_8f2read
mistral_ai.get_agent
Get Agent✓ allowed · 164ms
09:41:09 · claude · u_8f2write
mistral_ai.create_ocr
Create OCR✓ approved · approved by user
09:41:12 · claude · u_8f2destructive
mistral_ai.delete_file
Delete File✕ blocked · policy: deny
EVERY MISTRAL AI CALL, ON THE RECORD
01 · USE CASES

What agents do in Mistral AI.

01

Generate a chat completion

Send a prompt to a Mistral model and get a completion for a drafting task.

02

Run OCR on a document

Extract text from a document image using Mistral's OCR capability, returning structured content for further use.

03

Approve before deleting a library

Require approval before deleting a document library, since it removes every indexed file in it.

02 · ACTIONS

105 Mistral AI actions, graded by risk.

Every Mistral AI action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.

read

69

Look things up. Allowed by default.

  • mistral_ai.get_agent
    Get Agent
  • mistral_ai.get_judge
    Get Judge
  • mistral_ai.get_model
    Get Model
  • mistral_ai.get_voice
    Get Voice
  • mistral_ai.list_files
    List Files
  • mistral_ai.get_dataset
    Get dataset
  • mistral_ai.get_library
    Get library
  • mistral_ai.list_agents
    List Agents
  • mistral_ai.list_judges
    List Judges
  • mistral_ai.list_models
    List Models
  • mistral_ai.list_voices
    List Voices
  • mistral_ai.get_campaign
    Get campaign
  • mistral_ai.get_workflow
    Get Workflow
  • mistral_ai.get_batch_job
    Get Batch Job
  • mistral_ai.get_connector
    Get Connector
  • mistral_ai.list_datasets
    List datasets
+ 53 MORE

write

29

Create and change things. Allow, or ask the user first.

  • mistral_ai.create_ocr
    Create OCR
  • mistral_ai.create_agent
    Create Agent
  • mistral_ai.update_agent
    Update Agent
  • mistral_ai.create_speech
    Create Speech
  • mistral_ai.create_library
    Create library
  • mistral_ai.update_library
    Update library
  • mistral_ai.create_batch_job
    Create Batch Job
  • mistral_ai.create_embeddings
    Create Embeddings
  • mistral_ai.create_moderation
    Create Moderation
  • mistral_ai.create_library_share
    Create library share
  • mistral_ai.update_agent_version
    Update agent version
  • mistral_ai.create_fim_completion
    Create FIM Completion
  • mistral_ai.create_chat_completion
    Create Chat Completion
  • mistral_ai.create_chat_moderation
    Create Chat Moderation
  • mistral_ai.update_fine_tuned_model
    Update Fine-Tuned Model
  • mistral_ai.update_library_document
    Update library document
+ 13 MORE

destructive

7

Delete, cancel or archive. Ask first, or deny outright.

  • mistral_ai.delete_file
    Delete File
  • mistral_ai.delete_agent
    Delete agent
  • mistral_ai.delete_library
    Delete library
  • mistral_ai.delete_conversation
    Delete Conversation
  • mistral_ai.delete_library_share
    Delete library share
  • mistral_ai.delete_library_document
    Delete library document
  • mistral_ai.archive_fine_tuned_model
    Archive Fine-Tuned Model
03 · HOW IT WORKS

Mistral AI in three steps.

  1. 01Your users connect Mistral AIThey add their Mistral AI api key on Arc0 Connect, under your brand. It goes straight into the vault.
  2. 02You set the rulesReads run, writes like “create OCR” can wait for the user, and “delete File” can be denied outright.
  3. 03Any agent can actYour agent calls Mistral AI through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
POLICY.TS
await arc0.policies.set('mistral_ai', {
  read: 'allow',
  write: 'ask',        // create_ocr
  destructive: 'deny',  // delete_file
})

# Claude Code: the same connection, one URL
$ claude mcp add --transport http arc0 \
    https://mcp.arc0.ai/u/u_8f2
04 · AUTH AND DATA

How Mistral AI connects.

Users add their Mistral AI api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.

The same Mistral AI connection serves your agent over MCP and your own backend over REST and the proxy, so a user connects once. How Arc0 handles credentials →

AUTH
API key
CREDENTIALS
Per-tenant encrypted vault
MODEL SEES
Results only, never credentials
AUDIT LOG
Every call, on every plan
05 · WORKS WITH

Use Mistral AI from any agent.

Claude
ChatGPT
Cursor
Codex
VS Code
OpenAI Agents SDK
Claude Agent SDK
Vercel AI SDK
Mastra
LangGraph
07 · FAQ

Mistral AI and Arc0, answered.

Q01

Can I use Mistral AI with Claude, ChatGPT or Cursor?

Yes. Connect Mistral AI to Arc0 once, then add your Arc0 MCP URL to Claude, ChatGPT, Cursor, Claude Code or any other remote-MCP client. Each assistant only gets the Mistral AI actions you allow.

Q02

How do users connect Mistral AI?

Users add their Mistral AI api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.

Q03

Which Mistral AI actions can my agent take?

105 in total: 69 read, 29 write and 7 destructive, such as “create OCR”. Your policies decide which of them each agent may call.

Q04

Can I stop my agent from deleting things in Mistral AI?

Yes. Actions like “delete File” are graded destructive. Set destructive actions to deny, or to ask so the user approves each one, and blocked calls still show up on the audit log.

Q05

Can my own backend call Mistral AI too?

Yes. The same Mistral AI connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.

Get started

Plug Mistral AI into your agent.

Your users connect Mistral AI once, under your brand. Your agent gets 105 actions behind your policies, with every call on the record.

Free to build · MCP + REST · Audit log on every plan