Replyz / CLI

Email for your agent, from the terminal.

The Replyz CLI is how agents on your machine get email: one command to sign up, one to create an inbox, one to wait for a code. It is what the Replyz skill runs inside OpenClaw, Hermes and Claude Code.

Get early access On GitHub

How do you run it?

No install needed. Run it with npx:

npx -y @replyz/cli signup --name my-agent

Save the printed key as REPLYZ_API_KEY in the environment, and send the claim link to your human.

Which commands are there?

CommandWhat it does
signup --nameCreates an account, the API key (shown once) and a claim link for the human
statusLimits, usage and whether the account is claimed
inbox create, inbox lsCreate and list inboxes
messages ls, read <id>List and read messages
wait --from --otp --timeout --sinceWait for a matching email and return the code or link
tell-human --subjectEmail your human, for approval or to report back
mcp installSet up the Replyz MCP server in your MCP client

Every command accepts --json.

Why a CLI for agents?

Remote agents such as Claude use the MCP server instead.

Where is it used?

OpenClaw, Hermes and Claude Code, through the Replyz skill.

Frequently asked questions

How do I install the Replyz CLI?

You don't need to install it: run it with npx -y @replyz/cli. Agents and the Replyz skill call it that way, pinned to a version.

Why a CLI and not just MCP?

A CLI call is one short line, uses little context and reads the API key from the environment, so the model never sees it. Agents on your machine (Claude Code, OpenClaw, Hermes) work faster with it. Remote agents use the MCP server.

Does the CLI output JSON?

Yes. Every command accepts --json, so agents can parse the result.

How does an agent sign up?

With replyz signup --name. The CLI solves the sign-up proof of work, prints the API key once and a claim link for the human.

Give your agent an inbox.

Replyz is in private early access. Tell us which agent you run.

Get early access