Replyz / OpenClaw

Give OpenClaw its own email inbox.

Your OpenClaw agent gets an address of its own in about a minute. It signs up for services, reads verification codes and links, handles replies, and asks you before anything is sent. You stop lending it your Gmail.

Get early access The skill on GitHub

How do you give OpenClaw an email inbox?

Add the Replyz skill to OpenClaw (from ClawHub, or the skill on GitHub).
Ask it to get an inbox. The skill runs the Replyz CLI, which signs up and creates the first inbox:

npx -y @replyz/cli signup --name my-openclaw
npx -y @replyz/cli inbox create --name signups

Claim the account. OpenClaw sends you a link. One click ties the account to you and lifts the sandbox limits.

How does OpenClaw authenticate to Replyz?

With an API key stored in OpenClaw's secrets as REPLYZ_API_KEY. The CLI reads it from the environment on each command, so the key is never written into the conversation and the model never sees it.

What can OpenClaw do with its own inbox?

Sign up for services

Fill in a sign-up form with its own address, wait for the confirmation and finish.

Read codes and links

replyz wait --from "*@service.example" --otp returns the code, even if the email came first.

Handle replies

Read supplier, booking or support emails and draft the answer for you.

Ask before sending

Email you, its human, for a decision. Nothing goes out without your approval.

What can you ask OpenClaw to do?

Message it from Telegram, WhatsApp or any chat app you use with OpenClaw:

"Sign me up for the newsletter tool and tell me when you're in."
"Get the login code from the bank email and finish the setup."
"Check the signups inbox every morning and summarise anything new."
"Reply to the supplier that Thursday works, but show me first."

Why not just connect Gmail to OpenClaw?

Scope. A Gmail connection gives the agent your whole personal inbox. Replyz gives it only its own mail.
Built for agents. Waiting for an email and pulling out a code are one command each, not IMAP polling and parsing.
Safer sending. The agent can't email arbitrary people from your address.
EU-hosted. Mail is stored in the European Union.

What is OpenClaw?

OpenClaw is an open-source personal AI agent that runs on your own machine and takes instructions from chat apps such as Telegram or WhatsApp. It is extended with skills: small Markdown files, shared on ClawHub, that teach it to use a tool. Replyz is one of those skills.

Frequently asked questions

Can OpenClaw have its own email address?

Yes. With the Replyz skill, OpenClaw creates its own inbox in about a minute, with an address like name@youreplyz.com. It receives email there instead of in your personal account.

Do I need to connect my Gmail to OpenClaw?

No. Connecting Gmail gives the agent your whole personal inbox. With Replyz it only sees email sent to its own address.

How does OpenClaw read a verification code?

It runs replyz wait with the sender and --otp. Replyz returns the code or link from the matching email, even if the email arrived before the command.

Where does the API key go?

In OpenClaw's secrets store as REPLYZ_API_KEY. The CLI reads it from the environment, so the key never enters the model's context.

Can OpenClaw send email from the inbox?

Only with your approval. It can email you, its human, and reply in threads someone started with it. Bulk, marketing and cold email are not allowed.

Is Replyz an MCP server or a skill?

Both. OpenClaw uses the skill and the CLI. Remote agents such as Claude on the web use the Replyz MCP server instead.