WhatsApp MCP server with per-chat permissions
Let Claude, ChatGPT, and Cursor read and reply to WhatsApp, only in the chats you choose. Every other conversation stays invisible to the AI.
100% Free· Open source · Self-host or hosted
Example of the Claude chat app using the Automata WhatsApp MCP server. The user asks Claude to catch up on the Acme project WhatsApp group and confirm Friday's delivery. Claude lists the chats in scope, reads 48 messages from Acme project, is refused when it tries to read the Family chat because it is outside the token's scope, then sends a confirmation to Acme project and summarizes the conversation.
The basics
What is a WhatsApp MCP server?
A WhatsApp MCP server connects your WhatsApp account to AI assistants through the Model Context Protocol. It gives tools like Claude and ChatGPT a small set of actions, such as listing chats, reading and searching messages, and sending replies, so they can work with your conversations instead of you copying and pasting them.
Automata is a free, open-source WhatsApp MCP that adds the missing piece: a permission check before every tool call, so the AI only sees the chats you approve.
Why scope matters
Most WhatsApp MCPs hand over your whole account.
Popular open-source servers give the AI every chat, every group, and the ability to message anyone. Automata enforces access chat by chat.
| Capability | Typical WhatsApp MCP | Automata |
|---|---|---|
| What the AI can read | Every chat and group on the account | Only the chats you allow on the token |
| Sending messages | Any contact or group | Only chats with write access, recipient verified first |
| Per-chat permission levels | None | Read, write, or no access for each chat |
| Revoking access | Unpair the device or stop the server | Revoke one token; other clients keep working |
| Remote clients (Claude.ai, ChatGPT) | Usually local stdio only | Streamable HTTP endpoint with bearer tokens |
Token scope
WhatsApp chats
send_message · search_messages
- Read + write
Design team
WhatsApp group
- Read
Investor updates
WhatsApp group
- Blocked
Family
WhatsApp group
PostgreSQL tables
query · execute
- Read
public.orders
PostgreSQL table
- Read + write
public.inventory
PostgreSQL table
- Blocked
private.payroll
PostgreSQL table
Out-of-scope call refused
private.payroll · permission_denied
Reference
WhatsApp MCP tools
Five tools, each limited to the chats on the token. Disabled tools are never registered, so the AI cannot even try them.
get-connection-statusreadChecks that the paired WhatsApp account is connected before anything else runs.
list-chatsreadLists only the private chats and groups the token is allowed to see.
read-messagesreadPages through message history for one allowed chat, newest first.
search-messagesreadFull-text search across the allowed chats, never the whole account.
send-text-messagewriteSends a text message to a chat with write access after verifying the recipient.
Full parameters are in the MCP tools reference.
Setup
Connect WhatsApp to Claude in three steps
- 1
Pair WhatsApp with a QR code
Deploy Automata on Railway or your own server, create a WhatsApp connection, and scan the QR code from WhatsApp → Settings → Linked devices.
- 2
Choose the chats the AI can use
Create a connector token and set each chat to read, write, or no access. Only the tools you enable are registered.
- 3
Add the MCP URL to your client
Paste the endpoint into Claude, ChatGPT, Cursor, or any MCP client. Start with get-connection-status to confirm it works.
Claude Code
claude mcp add --transport http automata \
https://your-automata.example.com/mcp/<connector-token>In Claude.ai or Claude Desktop, open Settings → Connectors → Add custom connector and paste the same URL. Clients that prefer headers can use /mcp with Authorization: Bearer. See all clients.
Use cases
What you can ask once WhatsApp is connected
- “Summarize what Client A asked for in WhatsApp this week.”
- “Find the address Maria sent in the Suppliers group.”
- “Draft a reply in the Design team group confirming the new deadline.”
- “List every open question in the Acme project chat since Monday.”
FAQ
WhatsApp MCP questions
What is a WhatsApp MCP server?
A WhatsApp MCP server exposes your WhatsApp chats to AI assistants through the Model Context Protocol, so tools like Claude or ChatGPT can list chats, read and search messages, and send replies on your behalf.
Is the Automata WhatsApp MCP free?
Yes. Automata is MIT licensed and free to self-host, including a one-click Railway template. A hosted version is available by invitation.
Does it use the official WhatsApp Business API?
No. Automata pairs a regular WhatsApp account as a linked device through Evolution API, the same way WhatsApp Web works. No Meta business verification is required.
Can my WhatsApp number get banned?
Unofficial clients always carry some risk, especially with repeated pairing or unsolicited messages. Use a spare number while testing and avoid bulk messaging.
Can the AI read WhatsApp groups?
Yes. Groups are included alongside private chats, and each group can be allowed or blocked individually on the token.
Does it download photos, voice notes, or other media?
No. Messages carry their text or caption only. Media files are not downloaded or sent to the AI.
Which AI clients work with it?
Any client that supports remote MCP servers over HTTP, including Claude (web, desktop, and Claude Code), ChatGPT connectors, Cursor, and custom agents.
Should I self-host or use the hosted version?
Self-host if you want message data on infrastructure you control. Choose hosted if you want the same per-chat permissions without operating Evolution API, Postgres, and Redis yourself.
Give your AI WhatsApp, not your whole WhatsApp.
Deploy the free WhatsApp MCP server in minutes. The same token model also covers Telegram and PostgreSQL.