Clappie — Turn Claude Code into a 24/7 Personal Agent

7 min read Original article ↗

▄█▄ █████ ▀▀▀▀▀

▄██▄ █████▄ ▀▀▀▀▀▀▀

▄███▄ ▄██████▄ ███████████ ▀▀▀▀▀▀▀▀▀▀▀▀▀

▄█▄ █████ ▀▀▀▀▀

▄██▄ █████▄ ▀▀▀▀▀▀▀

▄████▄ ▄████████▄ ▄█████████████ ██████████████████ ▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀

▄█▄ █████ ▀▀▀▀▀

▄██▄ █████▄ ▀▀▀▀▀▀▀

▄█▄ █████ ▀▀▀▀▀

▄██▄ █████▄ ▀▀▀▀▀▀▀

˚ ° ˚ ° + ° · · ° · ˚ ° * + * + ˚ ° · + ° · * * ˚ ° ° * + ·


     
     
      
     
     
  

    
    

     
     

     
     
     
     

  
     

$ git clone https://github.com/whatnickcodes/clappie Copied!

Create, Manage, And Access Your Claude Code Terminals like they're 24/7 Mobile Personal Agents

Tutorial view source on github

Chat to Your Terminal

Skip the complex personal agent frameworks — just spin up as many simple Claude Code terminals as you need, on demand.

check my emails 14:32 ✓✓

3 unread:
• Sarah — API rate limits
• Mom — dinner Sunday?
• Vercel — deploy failed 14:32

reply to mom, say yes, skip creating a chore 14:33 ✓✓

Done! Replied to Mom ✓ 14:33

Message

~/clappie — claude

$ claude

╭─ Claude Code ─────────────────────╮

[sidekick] telegram: "check my emails"

Reading email inbox...

Found 3 unread emails

Sending summary to Telegram

Summary sent

telegram: "reply to mom, say yes, skip creating a chore"

Drafting reply to Mom...

Sent via email API

✓ Sidekick complete

Slack — Acme Inc

# engineering

M

Marcus 10:41 AM

can someone check why API latency spiked? seeing 2s+ on the dashboard

🐕

Clappie APP 10:41 AM

Redis connection pool exhausted — restarted, latency is back to normal.

💬 2 replies

S

Sarah Chen 10:43 AM

@clappie review PR #847 when you get a chance

🐕

Clappie APP 10:44 AM

PR #847 reviewed — LGTM! Left comments on the auth middleware.

💬 1 reply

session 1 — #engineering

[sidekick] #engineering → marcus

Checking API metrics...

Redis pool exhausted — restarting

Posting fix to thread

Replied in #engineering

session 2 — PR review

[sidekick] #engineering → @sarah

Fetching PR #847 diff...

12 files changed — reviewing

LGTM, noting auth middleware

Review posted

Heartbeat (cron)

Schedule background tasks to scan your emails, triage notifications, send telegrams, draft replies — whatever you want, on autopilot.

📧 scanned 14 new emails, archived 9

✈️ telegram → mom: "dinner at 7, bringing salad"

✏️ drafted reply to Rachel re: board demo

📋 created chore: "email boss Q1 roadmap update"

✈️ telegram → you: "PR #847 needs your review"

📅 synced 4 calendar events, conflict at 2pm

💬 slack: resolved alert in #engineering

🐙 github: merged 2 dependabot PRs

✈️ telegram → you: "mom emailed about dinner"

🗑️ unsubscribed from 5 newsletters

🧹 cleaned 340MB from ~/Downloads

💾 backed up memory + synced recall/

Displays (TUI)

Interactive terminal UIs that render right inside your Claude Code pane — notifications, dashboards, forms, anything. Plus and most importantly a fun little ASCII dog for your Claude Code crab to hang out with.

Telegram — chat into your terminal from anywhere

Notifications display

"show me my notifications"

Email display

"open my email inbox"

Slack — threaded replies, each spawning its own terminal

Parties — AI swarm simulations with shared rules

Chores approval queue

"show me my pending chores"

Chore revision

"revise that email draft"

Heartbeat scheduler

"check my heartbeat status"

Background manager

"show background services"

Push/Pop Navigation Mouse/Touch Clicks Automatic Keyboard Shortcuts Dark/Light Theme UI Kit for Custom Buttons & Forms Live Data Two-Way Communication Pixel Art Graphics Zero Dependencies Fully Responsive

Parties (Bot Coordination)

Swarms made simple. Define a game in plain text, spawn autonomous AI agents that share a ledger, exchange private messages, form alliances, bluff, negotiate, and self-terminate.

Plain Text Games Shared Ledger Private DMs Dice & RNG Squads & Teams Any Model Mix Self-Terminate Negotiate & Bluff

Many Mobile Options

Use Claude Code like you do on desktop from anywhere

Use our easy SSH setup guide to get the Claude Code terminal on your phone to have ALL the features you would have on your desktop.

Discord

WhatsApp

Telegram

Instagram

Slack

Signal

iMessage

Reddit

Twitter

LinkedIn

YouTube

Gmail

Setup to talk to Clappie from any app you already use via webhook integration helper

~$ ssh clappie@home

Connected to macbook.tail1234.ts.net

Last login: Mon Feb 24 09:41

~$ claude

╭────────────────────╮

│ Claude Code v1.0.3 │

│ ~/clappie │

╰────────────────────╯

> check my emails

Reading inbox… 3 new

✓ Summarized & created chores

>

Piggy back off this setup for super easy SSH

Dead Simple Setup

A Single Skill / Dead Simple Setup
No complex agent frameworks

It's really just a super minimal codebase that relays and piggybacks you to Claude Code from anywhere. The code is basically 1 skill.

~/clappie

clappie/

├── .claude/skills/

│ ├── your-skill/ ← build your own

│ │ ├── SKILL.md

│ │ ├── webhook.json ← webhooks to your machine

│ │ ├── oauth.json ← instant OAuth helper

│ │ └── your-skill.js ← auto-extends clappie CLI

│ ├── clappie/ ← the whole engine

│ │ ├── clapps/ ← displays, sidekicks, heartbeat…

│ │ └── clappie.js

│ ├── telegram-bot/ ← message from anywhere

│ └── [+ 3 more: slack, github, whisper]

├── recall/

│ ├── memory/ ← .txt files

│ ├── settings/ ← .txt files

│ ├── oauth/ ← token storage

│ ├── files/ ← photos, voice, video

│ ├── parties/ ← AI swarm games

│ ├── logs/ ← .txt files

│ └── profile.txt ← synthesized user profile

├── chores/

│ ├── humans/ ← prepared AI drafts, you approve

│ │ ├── reply-cto-roadmap.txt ← .txt files

│ │ └── call-dad-birthday.txt ← .txt files

│ └── bots/ ← automated heartbeat tasks

│ ├───── clean-notifications.txt ← .txt files

│ └───── system-health.txt ← .txt files

├── notifications/

│ ├── dirty/ ← raw dumps from integrations

│ │ ├── gmail-sync.txt ← .txt files

│ │ └── slack-sync.json ← technically a .txt file

│ └── clean/ ← triaged, ready to review

│ ├───── 3-new-work-chores.txt ← .txt files

│ └───── 6-auto-archived-emails.txt ← .txt files

├── projects/ ← websites, apps, anything

├── .env ← API keys (gitignored)

├── .gitignore

└── CLAUDE.md ← that's it, that's the whole thing

Background

  • Easy Tailscale private VPN helper
  • Background cron runner
  • Sidekick control center

📁

.txt File Philosophy

  • Simple folder structure
  • Plain text settings & config
  • Claude reads & writes naturally

🔧

Skill Builder

  • Skip risky third-party skill dirs
  • Build your own integrations instead
  • OAuth & webhook managers baked in

CODED IN ENGLISH

This was my personal Claude Code setup being released as a project. It was "vibe coded" — the architecture is human, the code is robot. Built entirely through conversation with Claude Code. Beware of serious issues, bugs or vulnerabilities. But it was also built with love.

NOT AFFILIATED WITH ANTHROPIC

Clappie is an independent project designed to work with Claude Code. It is NOT created, endorsed, or supported by Anthropic. "Claude" and "Claude Code" are trademarks of Anthropic. This software is not guaranteed to work, may break at any time without notice from a Claude Code update or from the author simply not maintaining it, may degrade or interfere with your normal Claude Code experience, and comes with zero expectation of continued functionality or compatibility.

By using Clappie, you acknowledge that your use of Claude Code remains subject to Anthropic's Acceptable Use Policy and Terms of Service. Clappie provides no separate license, warranty, or terms of its own. All AI interactions are processed through your existing Claude Code subscription and are governed solely by your agreement with Anthropic. Clappie's author assumes no liability for any actions taken by Claude Code on your behalf, any costs incurred through API usage, or any consequences arising from the use of this software.

EXPERIMENTAL SOFTWARE - EXTREME RISK

This is autonomous AI software that takes actions with minimal human oversight. It should ONLY be used in highly controlled, sandboxed, monitored environments - never with real accounts, real data, or production systems. It is NOT SECURE, has NOT BEEN AUDITED, and may cause IRREVERSIBLE HARM to your data, systems, accounts, finances, or reputation. There is NO WARRANTY of any kind. The authors accept NO LIABILITY and NO RESPONSIBILITY for any consequences. By using this software you assume ALL RISK. If you are not fully comfortable with experimental autonomous software acting on your behalf, DO NOT USE THIS.