CronyClaw

A local-first agent runtime with voice + Skills + OpenClaw bridge, packaged as a modern desktop app.

English · Tiếng Việt · 中文

---

Demos

Demo 1 cronyclaw_demo_1.mp4

Demo 2 cronyclaw_demo_2.mp4

---

This monorepo contains two main components:

• cronyclaw-backend: Python backend (uv run run_server.py) + Docker Compose
• cronyclaw-desktop: Electron desktop app (React + TypeScript)

Project Structure

• cronyclaw-backend/ — backend service, API/WebSocket, Docker setup
• cronyclaw-desktop/ — desktop client app

Quick Start

1) Run Backend

Option A — Local with uv

cd cronyclaw-backend
uv sync
uv run run_server.py

Option B — Docker Compose

cd cronyclaw-backend
docker compose up --build

Backend endpoints:

• HTTP: http://127.0.0.1:12393
• WebSocket: ws://127.0.0.1:12393/client-ws

2) Run Desktop

cd cronyclaw-desktop
npm install
npm run dev

Build Desktop

cd cronyclaw-desktop
npm run build:win
# or: npm run build:mac
# or: npm run build:linux

OpenClaw Integration

CronyClaw can receive updates from OpenClaw in two ways:

1. HTTP skill (claw-notify) — Add an agent skill (like claw-notify-skill) so OpenClaw posts to http://127.0.0.1:12393/claw-notify. Fast to set up on one machine; behavior depends on the agent following the skill.
2. Gateway WebSocket bridge — Configure OPENCLAW_GATEWAY_URL / OPENCLAW_GATEWAY_TOKEN (or conf.yaml) so the backend connects to the OpenClaw gateway, completes token-based connect, and may require openclaw devices approve <request_id> for pairing. More setup, closer to a stable event stream.

Detailed guidelines

• Backend — In the backend README, open the section Connecting CronyClaw to OpenClaw for architecture, both integration paths, conf.yaml / conf.example.yaml, env vars, device approval, and cache paths: English, Vietnamese, Chinese.
• Desktop (frontend) — In the desktop README, open Connecting OpenClaw (Settings) for Settings → General, Base URL, OpenClaw bridge URL/token, and Connect & Check Bridge: English, Vietnamese, Chinese.

Acknowledgements

CronyClaw is built by inheriting from and extending Open-LLM-VTuber. Thanks to that project and its contributors for the foundation.