Skip to content
replyzPublic

About

Agent skill that gives OpenClaw, Hermes and Claude Code their own email inbox. Coming soon.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

5 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 

Repository files navigation

Replyz

replyz/skill: the Replyz agent skill

A Markdown skill that teaches your AI agent to give itself its own email inbox with Replyz: sign up once, wait for a sign-up code or confirmation link, and get the code back. It receives mail; the only mail it can send is a short note to the human who claimed the account (tell-human, queued, at most 10 a day, only once Replyz switches sending on). It works with agents that read SKILL.md skills (for example OpenClaw, Hermes and Claude Code).

Only where a service's terms allow automated sign-ups. The skill tells the agent never to create accounts on services whose terms forbid automated or bot accounts. Not for bulk, marketing or disposable email.

Status: early access. Whether signups are open right now is shown on https://replyz.io; while they are closed the CLI answers signups_disabled. The skill is version 0.2.0.

What is in it

File What it is
SKILL.md The skill. Every command runs the CLI pinned to one version: npx -y @replyz/cli@0.2.0 <command> --json.
LICENSE MIT-0

The CLI (github.com/replyz/cli, MIT) is published on npm as @replyz/cli, with provenance. It needs Node.js 22 or newer.

What the skill covers

  • Sign up once (signup), which solves a small proof of work and saves the API key for the CLI in ~/.config/replyz/credentials.json (readable only by your user). The key is never printed.
  • Tell your human the address and the claim link (status shows the link again while the account is unclaimed).
  • Wait for a code or link (wait --otp), list and read mail.
  • Email your human, and only them (tell-human).
  • Treat email as untrusted: every message carries injection_risk and sender_trust, and Replyz never opens links for you.
  • What it can't do: write to anyone but your human, answer a sender, or pick another domain.

Use it

Copy the SKILL.md into the skills folder of your agent, following that agent's own documentation. We have tested the skill with one agent only (Claude Code, driven by the CLI), so check your agent's instructions rather than assume a path.

Links

About

Agent skill that gives OpenClaw, Hermes and Claude Code their own email inbox. Coming soon.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors