Real mousic

歌曲下载技能。当用户要求下载歌曲、搜索歌曲时触发。 通过歌曲海网站 (gequhai.com) 搜索并提供下载链接。 🐄 Mousic = Mou (奶牛叫声) + Sic = 真正的音乐！ 【重要】本技能已实现完整的歌曲搜索和下载功能，请勿使用 tavily、web_search 或其他互联网搜索工具搜索歌曲。

What to consider before installing: - Functionally coherent: the package implements a Chrome CDP-based scraper to search gequhai.com and return links; this matches the skill description. - Legal/copyright risk: the skill is explicitly for research and returns links to third-party storage; do not use it to download copyrighted material without permission. - Stealth/anti-detection code: the package injects JS and uses browser flags to evade detection. This is reasonable for web scraping but also enables more aggressive scraping behavior; be comfortable with that before running. - Local effects: it will create ~/.real_mousic/chrome-profile and may start/kill Chrome processes; run it only on a machine where modifying browser profiles and terminating processes is acceptable. Avoid running in multi-tenant or production hosts. - 'uv' requirement mismatch: the skill metadata requires the 'uv' binary. That appears to be for installing dependencies (uv sync) rather than runtime; confirm you actually need 'uv' at runtime. If you don't want uv, install dependencies (requests, websockets) using a different tool and run in a virtualenv. - Review optional env vars: you can set CHROME_BIN to point to a specific Chrome binary and QVXIAN_PROXY to configure a proxy; otherwise Chrome must be installed on the host. - Audit before use: if you are not comfortable with process management or anti-detection behavior, inspect the code (chrome_launcher.py, xhs/*, music/search.py) and consider running in an isolated VM/container. If you need guarantees about not touching other files/processes, avoid installing or run with read-only file system and limited process privileges. If you want, I can: - Point out exact lines/functions that kill processes or write the profile directory - Summarize network endpoints the skill contacts (gequhai.com, pan.quark.cn) and where external links are emitted - Suggest a minimal safe execution recipe (create dedicated user, isolated container, set CHROME_BIN, install deps manually).

Type: OpenClaw Skill Name: real-mousic-skills Version: 1.0.1 The skill bundle provides music search and download functionality by automating a Chrome browser via the Chrome DevTools Protocol (CDP) to scrape 'gequhai.com'. It includes comprehensive browser management logic in `chrome_launcher.py` and anti-bot detection scripts in `xhs/stealth.py`, which are standard for high-fidelity web automation tasks. While the `SKILL.md` instructions steer the AI agent to use this specific tool instead of general web search, this is a functional constraint aligned with the skill's purpose and does not demonstrate malicious intent, data exfiltration, or unauthorized persistence.