Annseraf
Permissions
Decide what each agent can do
Give one agent read-only access and another the right to draft replies. Sending stays off until you switch it on, agent by agent.
An Annseraf product
mcpmail: let AI agents use your email, on your terms.
Connect the mailbox you already have to Claude, ChatGPT, Cursor and other AI agents. You decide what each agent can do — read, draft, organize or send — and see everything it does.
What it does
Your mailbox, your rules, any agent
Visibility
Every action, on the record
See each tool call an agent makes on your mailbox: what it did, when, and whether it was allowed. Revoke any agent with one click.
Your mailbox
No new address, no migration
Fastmail, iCloud, Purelymail, Zoho, Yahoo, Gmail with an app password, or any IMAP server. Your mail stays where it is.
Any agent
Works with the AI you already use
Claude, ChatGPT, Cursor, VS Code and any client that speaks the Model Context Protocol, with setup steps for each.
Safety
Built for untrusted email
Email content is marked as untrusted for the AI, hidden text and tracking pixels are stripped, and agents are told to confirm sends and deletions with you.
Open source
The part that touches your mail is public
mcpmail runs on the open-source mcp-mail engine. Read the code, or run it yourself on your own server any time.
How it works
Three steps, a few minutes
- Connect your mailbox with an app password. We test the connection before saving it.
- Add an agent and pick what it may do. You get a link to paste into Claude, ChatGPT or your editor.
- Ask your agent about your email. Watch its activity, change its permissions or revoke it any time.
Start free
The Free plan gives you one mailbox and one agent, no card needed; paid plans add sending, more mailboxes and more agents. Prefer to run it yourself? The engine is open source.