manfredch/lyapunov/desktop-automation
Claude Code and agent skill by manfredch/lyapunov
Trust tier 3
- community
Source check pending
Checking that the SKILL.md file is still in the repository.
Use computer-use for local desktop applications (windows/clicks/keyboard/clipboard), or browser-use to open an isolated browser and visit web pages. Use for automate desktop app, click window, or browse web page requests; this does not operate the Lyapunov 3D scene itself.
Install
npx vskill@latest install manfredch/lyapunov/desktop-automationSKILL.md
Desktop and Browser Automation Rules
原生 ComputerUseLinux(使用当前发现的 mcp__computer-use-linux__* 工具)
- 先用
get_app_state按实际pid、window_id或应用名读取新状态;用list_windows/focused_window核对目标。仅操作用户已授权的目标,优先最新element_index或唯一name/role语义选择器。 click、set_value、press_key后读取新状态和应用实际结果;返回ok:true本身不证明点击、输入或提交生效。要求 Enter 时需实际按键/提交读回,perform_action的语义 activate 只是另一种操作。- 用
screenshot或带截图的get_app_state看真实目标。截图坐标先除以返回的scale;窗口相对点击使用relative:true并带同一目标,勿把逻辑控件坐标当桌面像素。 - 焦点拒绝时先核对当前窗口;可用原
activate_window/screenshot正常恢复目标前台,再读 exact focus。系统 portal 需正常授权,首次提示及时处理;授权等待失败后读新状态再重试,勿绕权限或盲目重复输入,也勿仅凭 fallback 的 ydotool 错误猜测缺驱动。
browser-use (mcp__chrome-devtools-mcp__* tools)
- Each session has an isolated Chromium instance, leaving the user's existing browser/debug channels untouched.
- After opening a page, inspect actual DOM/snapshot content before answering. A load-event timeout does not establish an unloaded page; inspect the DOM.
- Manage only tabs opened by this session. Do not terminate browser processes or operate on other user windows.
- Screenshots require image-input support; prefer text/DOM paths.
File Boundary (ENV-04)
- Use bash/read/write and other file tools for task-workspace scripts, references, intermediate outputs, and exports.
- Product state must use domain tools (
scene_*/asset_*/sim_*/robot_*, etc.). Do not change product state by editing files on disk, and do not modify product source. Capabilities come from skill contracts and actual product-tool results.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingpackages/lyapunov-shell/skills/desktop-automation/SKILL.md
Popularity
- Repository stars
- 511
- Repository forks
- 39
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/manfredch/lyapunov/desktop-automation)