For AI agents
Give your AI agent its own email address
Your agent signs up for a service, then gets stuck at “check your email”. Give it an inbox on your own domain, such as agent@yourname.si: it waits for the code, follows the confirmation link and reads replies, through our MCP server.
Works with Claude Code, Cursor and other MCP clients. The first mailbox on a domain registered with us is free.
- YouSign up for the analytics tool we picked, with agent@studio-nova.si, and finish the setup.
- AgentSigning up with agent@studio-nova.si. They sent a verification code, so I'm waiting for it.
- Toolwait_for_message · from: “analytics” · 4 s → “Your verification code: 482 913”
- AgentCode entered, the account is confirmed. The welcome email has a setup link; I followed it and the site is connected.
What agents use it for
Sign-up codes
Verification codes and one-time passwords that services send by email, picked up the moment they arrive.
Confirmation links
“Confirm your email” links, returned as text with their address, so the agent can open them.
Replies from people
Your agent emailed a supplier from its own tools? It can read the answer when it comes.
Receipts and alerts
Invoices, order confirmations and notifications from the services your agent looks after.
How it works
Connect the agent
Create an access token with the mail:read permission and add our MCP server to your agent.
The agent prepares a mailbox
It picks an address on your domain and gives you a link. You accept the terms and create it with one click.
It reads its mail
The agent lists messages, reads them as text, or waits up to 50 seconds for a new one, such as a code.
Why an address on your own domain
- It is yours
- agent@yourname.si belongs to your domain, not to a shared service. Services see your name, and you keep the address if you change agents.
- A real mailbox
- It is a normal mailbox on our mail server in the EU. Open it yourself in webmail or any mail app whenever you want to see what arrived.
- One place
- Domain, DNS, mailboxes and the agent connection in one account, with one support team.
Built to be safe
- Receive-only: the agent can read mail but cannot send, move or delete it.
- Reading needs the mail:read permission. A token only has it if you tick it, and you can revoke the token at any time.
- A token only reaches your own mailboxes. Your dashboard shows its recent activity.
- Every message is marked as written by its sender, so the agent treats it as data, not as instructions.
- No tool creates or pays for a mailbox: you do that yourself, with one click.
Price
Free
The first mailbox on a domain registered with us.
€19 a year
Each further mailbox, or one on a domain registered elsewhere. VAT included.
No domain yet? A .si costs €19.99 a year. Find a name
Questions
- Can the agent send email?
- Not from this inbox. It is receive-only for now: sending from a shared mail server is how good addresses end up on spam lists, and we protect our customers' business mail first.
- Which agents work with it?
- Any agent that connects to remote MCP servers over HTTP with a custom header: Claude Code, Cursor and others. ChatGPT cannot connect yet, because it only supports OAuth sign-in.
- Can I read the mailbox myself?
- Yes. It is a normal mailbox with webmail and IMAP. Its password is in your dashboard, where you can also reset it.
- Where is the mail stored?
- On our own mail server in the EU. The agent reads it through our MCP server with your token; we do not pass it to anyone else.
- What about prompt injection?
- Anyone can send an email, so mail is untrusted. Every message the agent reads is marked as the sender's text, and the agent cannot send mail or spend money from it. Give the token only the permissions it needs.
- Can I use a domain registered elsewhere?
- Yes, after you prove it is yours with a DNS record. A mailbox on it costs the yearly price; the free one is for domains registered with us.
Give your agent an inbox
Create a token, connect your agent, and ask it to set up its own address.
Create a token