AI assistants like Claude and Cursor can already read your codebase, run commands, and browse the web. But one boring step still stops them cold: email verification. The moment a sign-up sends a confirmation code or a magic link, your agent hits a wall — it can't read your inbox. The Temp Mail MCP server removes that wall. It gives your AI assistant its own disposable inboxes, so it can create an address, read the incoming email, pull out the code, and keep going — on its own, in seconds.
Table of contents
- What is MCP?
- What your agent can do
- Set up in under a minute
- Example: a sign-up that finishes itself
- Example: hands-free testing
- Why it's so fast
- Security and privacy
- Frequently asked questions
What is MCP?
MCP — the Model Context Protocol — is an open standard that lets AI assistants securely call external tools, instead of you copy-pasting data in and out of a chat window. Our MCP server exposes temp-mail.io as a set of tools your agent can call: create an inbox, list messages, read a specific email, download an attachment, and more. If your client speaks MCP — Claude Code, Claude Desktop, Cursor, Codex, VS Code, and Gemini CLI all do — it can use Temp Mail.
What your agent can do
- Create and delete temporary email addresses on demand
- List the messages in an inbox and read any one of them
- Pull verification codes and magic links straight out of the email
- Download attachments and inspect the raw message source
- List available domains and check the current rate limit
Set up in under a minute
There's nothing to install or self-host — it's a hosted, remote streamable-HTTP server secured with OAuth. Point your client at it once and you're done.
Claude Code:
claude mcp add --transport http temp-mail https://api.temp-mail.io/v1/mcpCodex:
codex mcp add temp-mail --url https://api.temp-mail.io/v1/mcpCursor and other clients — add the endpoint to your MCP config:
{ "mcpServers": { "temp-mail": { "url": "https://api.temp-mail.io/v1/mcp" } } }The first time your agent uses the server it opens your browser to sign in and authorize — there's no API key to copy or paste. After that, tokens refresh automatically.
Example: a sign-up that finishes itself
Here's the part that feels like magic. Ask your assistant to sign up for something and let it handle the whole email dance:
You: "Create a temp-mail inbox, use it to sign up for that newsletter, then paste back the confirmation code."
Claude: "Done. I createdk7m2p9@…, submitted the form, waited for the email, and the confirmation code was 481920 — the subscription is confirmed."
No polling script, no IMAP setup, no copy-paste. Under the hood the agent called create_email, then list_messages, then get_message, and read the code out of the body itself. What used to be a manual interruption is now a single sentence.
Example: hands-free testing
The same thing scales to QA. If you build or test sign-up flows, your agent can spin up a fresh inbox per run, complete the registration, grab the one-time code, verify the flow works, and throw the inbox away. What used to need a dedicated email-testing service and a pile of glue code becomes:
"Register five test accounts on staging, confirm each one from its own temp inbox, and tell me which succeeded."
Because every inbox is disposable and created on demand, there's no shared state to clean up and nothing tying the test accounts back to a real address.
Why it's so fast
- No integration code — the agent calls the tools directly; you don't write a client.
- No polling loops or IMAP setup — ask for messages when you need them.
- Nothing to run locally — it's a hosted endpoint, not a process you babysit.
- Connect once — OAuth in the browser, then tokens refresh on their own.
Security and privacy
Access is scoped to your temp-mail.io account through OAuth 2.0 — there's no long-lived API key sitting in a config file, and you can revoke the connection at any time. The inboxes themselves are disposable by design, so the data your agent touches is short-lived. It's automation without a permanent paper trail. And when you need a lasting private address instead of a throwaway one — say, for an account you'll keep — a forwarding alias from Alias Email is the better fit.
💡 Prefer raw REST? The same capabilities are available through the Temp Mail API.
Frequently asked questions
Do I need an API key?
No. Authentication is OAuth 2.0 in your browser, and your MCP client stores and refreshes the token for you. There's nothing to copy or paste.
Which clients are supported?
Any MCP-compatible client — Claude Code and Claude Desktop, Cursor, Codex, VS Code, Gemini CLI, Amp, and more. Per-client setup snippets are in the MCP documentation.
What's the endpoint?
It's a remote streamable-HTTP server at https://api.temp-mail.io/v1/mcp. Point your client there and authorize in the browser.
Can I use it in automated tests or CI?
Yes — creating inboxes and reading verification codes programmatically is a core use case. Your agent can register and confirm accounts as part of a test run.
Do I need a subscription?
Yes. The MCP server is a Temp Mail Premium feature, so you'll need an active Premium plan to connect. Premium starts at $2.99/month (or $29.90/year), and you sign in with that account when your client authorizes.
Bring your inbox into the agent loop
Email verification has been the last manual step in a lot of otherwise-automated workflows. With the Temp Mail MCP server, your AI assistant handles it end to end — create an inbox, read the code, move on. Add the server to your client in one command and try it at temp-mail.io; full setup details live in the MCP documentation.
