give your openclaw agent an email in 60 seconds

Your OpenClaw agent can browse the web, run shell commands, and call APIs. But the moment it needs to send or receive an email, you're stuck wiring up OAuth scopes, IMAP connections, and app passwords against your personal Gmail. There's a better way.

LobsterMail lets your agent pinch its own inbox — a real email address it controls, isolated from your personal mail. And with the LobsterMail skill for OpenClaw, the whole setup takes about 60 seconds.

Let's walk through it.

what you'll need#

• A running OpenClaw instance (local or hosted)
• Node 22 or later (required by OpenClaw's runtime)
• An OpenClaw workspace with at least one configured AI provider

If you haven't set up OpenClaw yet, the official installer handles everything:

curl -fsSL https://openclaw.ai/install.sh | bash

Then run through the onboarding wizard:

openclaw onboard --install-daemon

Once your lobster is alive and responding to messages, you're ready to give it email.

step 1: install the lobstermail skill#

OpenClaw's skill system makes adding new capabilities dead simple. Skills are modular — install one, and your agent immediately knows how to use it.

Install the LobsterMail skill from ClawHub:

clawhub install lobstermail

Or if you prefer npm directly:

npx openclawskill install lobstermail

That pulls in the LobsterMail SDK and registers the skill with your OpenClaw gateway. Your agent can now provision and manage email inboxes on its own.

step 2: let your agent hatch an inbox#

Here's where it gets fun. You don't create the inbox. Your agent does.

Just tell it in natural language:

Hey, get yourself a LobsterMail inbox so you can handle email.

Behind the scenes, the agent calls the provisioning function from the LobsterMail SDK. If you're building a custom integration or want to understand what's happening under the hood, here's the equivalent code:

import { LobsterMail } from "@lobstermail/sdk";

const client = new LobsterMail();

const inbox = await client.provision({
  name: "my-openclaw-agent",
  webhookUrl: "https://your-openclaw-instance.local/webhook/lobstermail",
});

console.log(inbox.address);
// → my-openclaw-agent@getlobstermail.com

One function call. That's it. No signup form, no email verification, no password to store. The agent hatches into the reef with a working address and is ready to receive mail immediately.

step 3: receive the first email#

Send a test email to your agent's new address from any email client. If you set up the webhook URL during provisioning, your OpenClaw instance receives the message instantly. The agent sees something like this in its context:

{
  "event": "email.received",
  "from": "you@gmail.com",
  "to": "my-openclaw-agent@getlobstermail.com",
  "subject": "Testing your new inbox",
  "body": "Hey agent, can you read this?",
  "receivedAt": "2026-02-24T10:30:00Z"
}

Your agent can now read it, respond to it, or route it however you've configured. No polling loop required (though LobsterMail supports that too if webhooks aren't your thing).

step 4: send a reply#

On the free tier, your agent can receive emails immediately. To unlock sending, verify via an X post or credit card — it's free, and gives you 10 sends per day. For higher volume, the Builder plan is $9/month.

Once sending is enabled, replying looks like this:

await client.send({
  from: "my-openclaw-agent@getlobstermail.com",
  to: "you@gmail.com",
  subject: "Re: Testing your new inbox",
  body: "Got it! My shell is live and I'm ready to work.",
});

Or in natural language, just tell your agent: "Reply to that email and confirm you received it." The skill handles the rest.

what to build next#

Now that your agent has its own email, the possibilities open up. A few ideas:

Customer support triage. Point your support address at the agent. It reads incoming messages, classifies urgency, drafts responses, and flags anything it can't handle. You can even use a custom domain so customers email support@yourcompany.com and your agent handles it.

Newsletter monitoring. Subscribe your agent to industry newsletters and RSS-to-email feeds. It reads, summarizes, and sends you a digest. All through its own inbox on the reef.

Multi-agent coordination. If you're running multiple OpenClaw instances, each agent gets its own address. They can pass tasks, share results, and collaborate through structured email threads. Check out what agents do with email for more patterns.

Automated follow-ups. Your agent sends a proposal, then checks its inbox for replies. If nothing comes back in 48 hours, it sends a follow-up. No cron jobs, no manual checking.

Why not just connect gmail?#

We get this question a lot. The short answer: security and separation.

When your OpenClaw agent connects to your Gmail, it has access to every message in your inbox. Bank statements, medical records, private conversations — all sitting in the agent's context window. A prompt injection attack embedded in a malicious email could hijack your agent's behavior with access to all of that data.

With LobsterMail, the agent's shell is isolated. If something goes wrong, the blast radius is limited to the agent's inbox. Your personal email stays untouched. For more on this, read what is agent email and why does it matter.

60 seconds, recap#

1. Install the skill: clawhub install lobstermail
2. Tell your agent to grab an inbox (or call client.provision())
3. Send it a test email
4. Watch it arrive

That's the whole thing. Your OpenClaw agent now has its own communication channel, separate from your personal email, ready to interact with the outside world.

The OpenClaw community — 140,000+ GitHub stars and over 5,700 community skills on ClawHub — is building the infrastructure for agents that actually do things. Email is one of the most universal protocols on the internet. Giving your agent its own address is one of the highest-leverage things you can do.

Give your agent its own email. Get started with LobsterMail — it's free.