666-gy/yan-agent/computer-use
Claude Code and agent skill by 666-gy/yan-agent
- community
Checking that the SKILL.md file is still in the repository.
Control Windows apps from Yan Agent. Use for desktop application tasks that need window discovery, screenshots, accessibility controls, clicks, typing, scrolling, and verification.
Install
npx vskill@latest install 666-gy/yan-agent/computer-useSKILL.md
Computer Use
Activate the installed yan-computer-use plugin with plugins_activate when needed, then use the tools and schemas returned for this session. If the plugin or this skill is disabled, stop and explain the missing capability; do not create another desktop driver or bypass the plugin with shell-based input automation.
Find the exact application and window. Read fresh get_window_state before interacting. Prefer actionable element tokens when the accessibility tree exposes the target. New snapshots invalidate previous tokens. If a control is missing, use coordinates from that exact window screenshot; the driver handles scaling from screenshot pixels to screen coordinates. Do not manually scale them twice.
Prefer background delivery. A successful delivery with effect: unverifiable does not prove the app responded: inspect fresh state before deciding what to do next. When background input has no effect, choose foreground delivery when authorized. A refusal does not authorize escalation. For an unchanged loading panel, avoid repeatedly taking identical screenshots without changing the plan; check focus, the target control, or whether a second interaction is appropriate.
Use the user's configured image-input capability. Text-only models can act on the accessibility tree; use existing yan_media_read_image with saved screenshot paths for visual interpretation when its relay is configured. Do not add a vision backend or infer multimodal support from the model name.
Before sending a message, verify the recipient and composed text. After any action, check the relevant resulting state. Never repeat a delivered send unless fresh state shows it did not succeed. Describe what the evidence establishes: a playback UI is evidence of a playing state, and a sent message bubble is evidence of client-side sending, not proof that every recipient read it.
Ordinary web tasks continue to use Yan Built-in Browser. Use this skill when the requested workflow requires interacting with a Windows application.
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 repositoryCheckinglib/plugins/catalog/yan-computer-use/skills/computer-use/SKILL.md
Popularity
- Repository stars
- 111
- Repository forks
- 14
- 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/666-gy/yan-agent/computer-use)