sales@agent.acme.com, live in about 30 seconds

Give your agents a way to reach people.

Every agent gets a real address on your domain, an inbox it can hold a conversation in, and a memory of everyone it has talked to. You watch all of it, and step in whenever you want.

or

Settings, then Connectors, then Add custom connector. Paste this and sign in.

https://connect.mcpmailer.com/mcp

No account needed first. You sign in when your app asks to connect, and that creates it.

Using something else?

Everything your assistant can do with email

Connect Claude, ChatGPT, Vercel AI SDK, or LangChain once, and it gains 67 email abilities: writing, replying, searching, waiting for an answer, remembering who it spoke to. Pick one to watch it run.

connect.mcpmailer.com/mcp
agent

Send from the agent mailbox. Replies carry generous quotas; cold sends are rate-limited.

What happens to a message

Agent mail is ordinary mail, so it has to survive the same checks ordinary mail does. Every send is authenticated, threaded against what came before, and metered by whether it is a reply or a cold approach.

Authenticate

SPF, DKIM, and DMARC signed on the way out and checked on the way in.

Thread

References and In-Reply-To carried over, so the reply lands in context.

Meter

Replies pass free. Cold sends draw down the quota you set.

Nothing your agent does is invisible

Every agent, every thread, every send and receive, in one place. Search the whole archive, read what a customer actually wrote back, and take over a conversation yourself when it needs a person.

An inbox is where it starts

An agent that only sends mail forgets everyone it has ever met. The workspace around the inbox is what makes the next conversation better than the last one.

Contacts

The address book your agents read before they reply. Anything an agent remembers about someone shows up here.

Notes

Shared context that outlives a conversation. You see every note; each agent sees only the ones you grant it, plus the ones it wrote itself.

Vault

Credentials your agents can use without you handing them over in plaintext. Encrypted in your browser with a key we never receive and cannot recover.

Webhooks

A signed POST to your endpoint the moment an agent receives mail, or when a send bounces or draws a complaint. No polling.

Suppressions

Addresses your agents will not email. Sends are rejected before they reach the provider, which is what keeps your sending reputation intact.

Connections

Apps you have authorized to act as your agents. Revoking one stops it immediately, and you see the same approval screen if it asks again.

Activity

What happened to your mail after it left, and anything we did about it. Bounces, complaints, unsubscribes, and every throttle or pause.

Usage

Thirty days across every agent, against the cold and reply quotas your tier allows, so a runaway agent is visible before it is expensive.

REST API

Everything the MCP server can do, over plain HTTP with the same key, for anything in your stack that does not speak MCP.

Your agents write from your domain

Add acme.com or agents.acme.com and we give you the settings to paste into your domain provider, then check them for you until they work. Nothing you already run is touched: existing mail keeps flowing exactly as it does today.

How domain setup works

See the spend before it surprises you

Thirty days of sends and receives across every agent, against the cold and reply quotas your tier allows. A single agent that has started talking too much is visible long before it shows up on an invoice.

See what a plan includes

Connect the client you already use

Pick yours and we show you exactly what to do. Claude, ChatGPT, and OpenClaw are a settings screen and a paste. The rest are a few lines in whatever you are building.

Settings, then Connectors, then Add custom connector. Paste this and sign in.

Connector URL

https://connect.mcpmailer.com/mcp
Full Claude guide

Terminal

openclaw mcp add mcpmailer --url https://connect.mcpmailer.com/mcp --transport streamable-http --auth oauth --timeout 320000
Full OpenClaw guide
~/.hermes/config.yaml
mcp_servers:
  mcpmailer:
    url: https://connect.mcpmailer.com/mcp
    auth: oauth
Full Hermes guide

Connector URL

https://connect.mcpmailer.com/mcp
Full ChatGPT guide
Agents SDK
await this.addMcpServer('mcpmailer', 'https://connect.mcpmailer.com/mcp', {
  transport: {
    type: 'streamable-http',
    headers: { Authorization: 'Bearer mmk_live_...' }
  }
});
Full Cloudflare Agents SDK guide
AI SDK (TypeScript)
import { createMCPClient } from '@ai-sdk/mcp';
import { generateText } from 'ai';

const mcp = await createMCPClient({
  transport: {
    type: 'http',
    url: 'https://connect.mcpmailer.com/mcp',
    headers: { Authorization: 'Bearer mmk_live_...' }
  }
});

const { text } = await generateText({
  model: 'openai/gpt-5.4',
  tools: await mcp.tools(),
  prompt: 'Email anna@customer.com her order status.'
});
Full Vercel AI SDK guide
LangChain (Python)
from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain.agents import create_agent

client = MultiServerMCPClient({
    "mcpmailer": {
        "transport": "streamable_http",
        "url": "https://connect.mcpmailer.com/mcp",
        "headers": {"Authorization": "Bearer mmk_live_..."},
    }
})

agent = create_agent("openai:gpt-4.1", await client.get_tools())
Full LangChain guide
Agents SDK (Python)
from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

async with MCPServerStreamableHttp(
    name="mcpmailer",
    params={
        "url": "https://connect.mcpmailer.com/mcp",
        "headers": {"Authorization": "Bearer mmk_live_..."},
    },
    cache_tools_list=True,
) as server:
    agent = Agent(name="Support", mcp_servers=[server])
    result = await Runner.run(agent, "Reply to the newest support email.")
Full OpenAI Agents SDK guide

Endpoint

https://connect.mcpmailer.com/mcp
Full n8n guide

What people put them to work on

The same tools, pointed at different jobs. Each one is a real subject line an agent on that job would send.

Questions

What people ask before they connect an agent: how addresses work, what happens to deliverability, and what you can see and stop.

What is MCPmailer?

MCPmailer is email infrastructure for AI agents. Each agent gets its own address, an inbox it can hold a threaded conversation in, and a memory of the people it has talked to. Your assistant reaches all of it over MCP, and you watch every message from a dashboard.

How is this different from an email API like Resend or Postmark?

A transactional email API delivers mail your product sends. MCPmailer is built for conversations: an agent receives replies, keeps threading in both directions, waits hours for an answer, and remembers who it was talking to. Sending is one tool out of many rather than the whole product.

Do I need to be a developer to use it?

No. If you use Claude, ChatGPT, or another MCP client, connecting MCPmailer is a settings screen and a paste, and your assistant does the rest. The REST API and the SDKs are there for people building software, not a requirement for everyone else.

What is MCP?

MCP, the Model Context Protocol, is the standard way an AI assistant connects to outside tools. A client that speaks MCP can use any MCP server, so connecting MCPmailer once gives your assistant every email ability it has without you writing an integration.

Can I use MCPmailer without MCP?

Yes. Everything the MCP server can do is available over a plain HTTP REST API with the same key, plus SDKs and a CLI. Agents that do not speak MCP, cron jobs, and back end services all use the same account, inboxes, and quotas.

How long does setup take?

An inbox on mcpmailer.email is live in about 30 seconds: sign up, create an agent, connect your client. Sending from your own domain takes as long as your DNS provider needs to publish three records, usually minutes.

Give your agent an address.

3,000 emails a month, free, no credit card.