liuzhijie443/yd-mclaw

This repository is a usage and deployment guide for MClaw, a tool for managing a mobile cloud drive container setup. It documents how to swap the underlying AI model, deploy the project, set up Cloudflare Tunnels for exposing services without a public IP, manage installed applications inside the container, and configure the container to start automatically after a reboot. The main audience is end users who already have a MClaw container running and want step by step instructions for common tasks. The guide walks through editing a configuration file to point the container at a different AI model by setting values for baseUrl, apiKey and model. It also explains how to download a deployment guide text file, fill in variables such as a gateway auth token, model connection details, and a Cloudflare Tunnel token, then hand that file to MClaw so it can carry out the deployment itself. Once deployed, the setup installs cloudflared for tunneling and a small SSH server called dropbear on port 2222, along with startup and watchdog scripts that keep services running and restart them automatically after a container reboot. The guide also covers configuring Cloudflare Tunnel routes so the OpenClaw web interface and SSH access are both reachable from outside the container, and shows the exact commands needed to tunnel an SSH connection to a local port and then connect with a normal ssh client. A troubleshooting section explains what to do if MClaw becomes unresponsive, how to ask it to manage installed programs, how to change configuration values after the fact, and how to install extra software. The README notes that resetting MClaw only wipes its own memory and configuration folder, not other data on the container. It also mentions optionally installing extra utilities afterward, such as a scheduled task panel, a file listing tool, and a proxy utility. The documentation is written entirely in Chinese and assumes the reader is already running this container setup through a specific hosting arrangement referenced in the text. There is no license or code structure described beyond the deployment steps themselves.