You decide what each agent can do
Give each of your AI tools its own name (like “My Cursor” or “Claude at work”). Choose whether it can send messages, read your chats, or set up new agents. Change this anytime.
WAMCP makes it easy and safe to let your own AI tools — Claude, Grok, Cursor, or any other — use your WhatsApp. You decide what each agent can do and can turn them off anytime.
Every agent action is audited. Self-chat MFA for sensitive operations.
Create named agents and decide exactly what each one can do.
Claude, Cursor, Grok and many others — the way they already work.
Health status, active agents and simple alerts at a glance.
Connect once, then manage everything from the simple portal.
Connect your WhatsApp once. Then decide exactly what your AI agents are allowed to do. Turn any of them on or off whenever you want.
Give each of your AI tools its own name (like “My Cursor” or “Claude at work”). Choose whether it can send messages, read your chats, or set up new agents. Change this anytime.
Claude, Grok, Cursor, Codex, and many others. Your agents connect the normal way they like to work. You don’t need to change how you use them.
Check if your WhatsApp connection is working, see which agents are active, and get alerts if something goes wrong. Everything stays private to you.
After connecting, open the simple settings page on your phone or computer. Create agents, change what they can do, or stop them — no complicated setup.
Every agent action is audited. Self-chat MFA for sensitive operations. Pairing code required for destructive actions. You always know who is connected and what they did.
Sync-aware, reliable WhatsApp connection that stays alive. Your agents get proper tools for notifications, search, replies, media, and groups — while you keep oversight.
Copy ready reactive loops, inbox triage, scheduled digests, memory recall and group tools from /skills. Full sequences, guardrails and audit transparency — not toy starters.
Pair your WhatsApp once. Then decide exactly what your AI agents are allowed to do. Turn any of them on or off whenever you want.
Give each of your AI tools its own name (like “My Cursor” or “Claude at work”). Choose whether it can send messages, read your chats, or help create new agents. Change this anytime.
Drop a small config into Cursor, Claude or Grok. Your agents get 33 real tools for notifications, search, replies, media and groups.
Use whatsapp_watch_notifications for reactive loops. Agents stay responsive and you see everything in the portal.
Every call is logged. Sensitive actions require self-chat confirmation from your own phone. Revoke any agent instantly.
Copy ready reactive loops, inbox triage, digests, memory recall and group tools from /skills.
After pairing you get a simple PWA link in self-chat. Name agents, toggle permissions, check health — nothing more.
You pair once. WAMCP keeps a reliable bridge. Your agents call standard MCP or REST. You control everything from the portal.
Use the connect page or ask your AI tool to start the pairing for you.
curl -X POST https://wamcp.up.railway.app/api/session/connect/kit \
-H "X-Tenant-Key: tnt_..." \
-d '{"phone_number":"5511999999999"}'
# poll until success
curl "https://wamcp.up.railway.app/api/session/connect/status?uuid=UUID"Paste the MCP block into Claude, Cursor, Grok or your agent host.
{
"mcpServers": {
"my-whatsapp": {
"url": "https://wamcp.up.railway.app/mcp",
"headers": {
"Authorization": "Bearer mcp_...",
"MCP-Protocol-Version": "2025-11-25"
}
}
}
}It lets your AI tools safely use your WhatsApp while you stay completely in control from a minimal portal.
No. You choose per agent what it can read or send. You can turn any of them off at any time.
No. Pair, name the agent in the portal, paste the MCP JSON into your tool. Start with the skills.
After sync completes you receive a self-chat link to the simple PWA settings page.