WSL Chrome CDP

自动检测并启动 Windows Chrome 调试模式，实现 WSL2 环境下对 Chrome 浏览器的无缝远程控制。

What to check before installing: - Inspect enable-browser.sh yourself (it is short and human-readable). It runs curl, unsets proxy vars, finds your Windows IP, and invokes PowerShell to start Chrome with --remote-debugging-port=9222 and a user-data-dir in your Windows profile. - The manifest claims permission to write ~/.openclaw/openclaw.json and to exec scripts/start-chrome-debug.bat, but the package does not include code that writes that file and the .bat is not present — ask the author to explain/remove the write permission or provide the missing file. - Starting Chrome with remote-debugging opens a local TCP port (9222). Make sure your firewall and network settings do not expose that port to untrusted networks. Only run this on machines you control. - If you are unsure, run the script manually from a terminal so you can see output and stop it; prefer manual invocation over enabling any automatic/unattended mode. - If you will allow the skill to edit ~/.openclaw/openclaw.json, back up that file first and verify any changes after installation. Overall: the functionality matches the description, but the manifest/permission mismatches warrant caution — treat this as a packaging quality issue that should be clarified before trusting automated installs.

Type: OpenClaw Skill Name: wsl-chrome-cdp Version: 1.0.0 The wsl-chrome-cdp skill is designed to facilitate communication between an OpenClaw agent in WSL2 and a Chrome browser instance on the Windows host via the Chrome DevTools Protocol (CDP). The core logic in enable-browser.sh uses PowerShell to launch Chrome with remote debugging enabled and a dedicated user data directory, which is a standard and legitimate approach for browser automation. While the skill requests write access to the agent's configuration file (~/.openclaw/openclaw.json) and executes host-side commands via PowerShell, these actions are transparently documented and directly support the stated purpose of the skill without any evidence of malicious intent or data exfiltration.