Agent Inbox runs on your own machine — the client starts it, so there is no endpoint to ping. Last commit 17 Jul 2026.
A real email inbox for AI agents: human-approved sends, untrusted-by-default inbound.
Today is the operative word: we check Agent Inbox every 15 minutes and re-read its code on every release. Watch it and you find out the day that stops being true.
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add inbox -- npx -y @clize/inbox
{
"mcpServers": {
"inbox": {
"args": [
"-y",
"@clize/inbox"
],
"command": "npx"
}
}
}
[mcp_servers.inbox]
command = "npx"
args = ["-y", "@clize/inbox"]
{
"mcpServers": {
"inbox": {
"args": [
"-y",
"@clize/inbox"
],
"command": "npx"
}
}
}
{
"mcpServers": {
"inbox": {
"args": [
"-y",
"@clize/inbox"
],
"command": "npx"
}
}
}
Email for AI agents — read inbox (IMAP), search, and send (SMTP).
Real email and SMS for AI agents — send mail, receive verification codes, drive a real inbox.
Check an email before an AI agent sends it: send / ask a human / block. Detects prompt injection.
Unified inbox MCP for WhatsApp, Telegram, Email, voice — read/send messages, search, AI agents.
Real email inboxes for AI agents: open an address, read the one-time code, send mail.
Email inboxes for AI agents: send, receive, reply, search, and manage threaded email over MCP.
Human approval gate for AI agents. Critical actions approved or rejected by a human via Telegram.
Stateful email for AI agents — read inboxes, reply in-thread, draft with approval.
Answers built from our own checks of this server.