MCP is live!

Updated 15 October 2026 to describe how the server works today: sign-in instead of pasting a key, 25 tools, and a listing in the official MCP Registry.

In order to increase your time-to-first-email, we've implemented MCP for your agentic workflows. Your assistant can add a domain, read back its DNS records, send a test email and draft a broadcast without you leaving the editor.

The server wraps our API, so every call goes through the same pipeline as a normal API request. DKIM signing, tracking, suppressions, rate limits and your plan's sending limits all apply exactly as they do for the API.

Connecting

The hosted server lives at https://mcp.eusend.dev/mcp. In Claude Code:

claude mcp add --transport http eusend https://mcp.eusend.dev/mcp

Then run /mcp and choose Authenticate. Your browser opens eusend, you sign in and approve the connection, and the client gets a token of its own. No API key is copied into a config file. Sign-in also works with clients that can't send a custom header, like Claude on the web: add the URL under Settings → Connectors.

The approval screen asks for three things:

• Workspace. The one workspace the connection can reach.
• Access. Full access, or sending only. A sending-only connection can send, reschedule and cancel email, and nothing else.
• Mode. Live, or test mode. In test mode, calls are accepted and tracked but nothing is delivered.

Connections are listed under Settings → Connected apps. Disconnecting one takes effect on its next request.

API keys still work if you'd rather use one. Pass the key as an Authorization: Bearer header, or run the server locally with npx -y @eusend_dev/mcp so the key never leaves your machine. The server is also listed in the official MCP Registry as dev.eusend/eusend, so clients that browse the registry can find it there.

What the assistant can do

Area	Tools	What they do
Emails	4 tools	send_email, list_emails, get_email, cancel_email: send transactional mail and inspect delivery.
Domains	4 tools	list_domains, get_domain, create_domain, verify_domain: add a sending domain, read back its DNS records and re-check them.
Audiences & contacts	8 tools	list/get/create/rename audiences, list and create contacts, list and create contact properties.
Topics	2 tools	list_topics, create_topic: the subscription topics contacts can opt in and out of.
Broadcasts	5 tools	list_broadcasts, get_broadcast, create_broadcast, test_broadcast, send_broadcast: draft, send yourself a test copy, then send for real.
Suppressions	2 tools	list_suppressions, create_suppression: see which addresses are blocked and why, and stop sending to one.

What it deliberately can't do

An agent will sometimes do the wrong thing confidently, so the tools are designed so that the wrong thing is hard to do:

• Nothing destructive. There are no tools for deleting domains, audiences, contacts or API keys.
• Broadcasts take two steps. create_broadcast only writes a draft, and send_broadcast is a separate call. test_broadcast sits in between and can only deliver to your own verified domains, so it can't reach a list by mistake.
• No un-suppressing. Adding a suppression stops mail. Removing one starts sending again to an address that bounced or complained, and doing that in bulk is how a sender wrecks its reputation.

Tools are half of it

The tools let an assistant act on your account. They don't stop it from writing your integration wrong. If you use Claude Code, the eusend plugin installs this server together with a skill that teaches Claude how eusend behaves: verified senders, retries that are safe to repeat, webhook signatures.

Setup for Cursor, Codex, Claude Desktop and other clients is on the MCP docs page.