clone
Skill17 starsChanged 11 days agoMITView on GitHub
Walk through any existing app (native Windows app with computer use, web app with a browser), screenshot every screen, menu, and dialog, write a screen index, and turn it into user stories. Use when the user wants to clone, rebuild, replace, or copy the design of an app, or says "screenshot every screen".
7 other projects use a version of this, 2 of them edited.
Adopted by
Most recently taken up by
- Stijnus/dsh-rev-engedited, first seen Nov 8, 2026
- deepaktechk/penpot-agenttheir own version, first seen Nov 1, 2026
- yowpi-tech/unbubbletheir own version, first seen Nov 1, 2026
- daveebbelaar/bamaedited, first seen Nov 1, 2026
- thenavidm/remastertheir own version, first seen Oct 20, 2026
- vinitjhawar/opencounciltheir own version, first seen May 17, 2026
Other versions7 projects use a version of this.
Word for word · 0
None found.
Edited · 2
- daveebbelaar/bama★ 1.7k67% same
- Stijnus/dsh-rev-eng★ 164% same
Wrote their own version · 5
- spotify/ads-agentic-tools★ 20same name
- yowpi-tech/unbubble★ 4same name
- vinitjhawar/opencouncil★ 1same name
- deepaktechk/penpot-agent★ 0same name
- thenavidm/remaster★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository23 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- agent-cliSkill
- csharp-winuiSkill
- improve-architectureSkill
- os-integrationsSkill
- performanceSkill
- app-iconSkill
- design-from-appSkill
- design-kitSkill
- design-systemSkill
- layout-gridSkill
- ux-designSkill
- windows-designSkill
- researchSkill
- user-storiesSkill
- build-and-verifySkill
- distributeSkill
- improve-loopSkill
- test-windows-appSkill
- app-briefSkill
- bawaSkill
- new-windows-appSkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 7 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 62
Every navigation item (the left pane or sidebar), top to bottom.
- Distinctive: line 64
Every view mode (list, board, calendar), sort menu, filter menu.
- Distinctive: line 122
Command map: every menu and command bar item, its keyboard shortcut, and its access key.
- Very specific: line 132
Call the Skill tool with user-stories.
“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Clone
Output: a folder of screenshots, one index that describes everything the app shows, and a backlog built from it.
docs/screens/
INDEX.md
01-inbox.png
02-today.png
...
docs/STORIES.md (from the user-stories skill)1. Pick the tool
| The app is | Use |
| A native Windows app (WinUI, WPF, WinForms, Win32), or Electron | Computer use in the Claude desktop app |
| A web app | A browser tool: the built-in browser, Claude in Chrome, or Playwright |
Turn on computer use in the Claude desktop app's settings. Ask for access to the reference app only. Browsers are view-only and terminals are click-only for computer use.
2. Agree on privacy
Tell the user you will see everything in the app. Suggest a demo account or sample data. Look, do not touch: open screens and menus, press Esc to close dialogs. Type, create, change, or delete only when the user says so.
3. Save real files
Computer use screenshots come to you downscaled, not to disk, and they show the whole desktop. Save each screen at full resolution, and only the app's window, with the script in this skill:
powershell -NoProfile -ExecutionPolicy Bypass -File "${CLAUDE_SKILL_DIR}/scripts/window-shot.ps1" "<process or title>" docs/screens/01-inbox.pngIt captures the largest visible window whose process name (without
.exe) or title matches, with PrintWindow, so a window behind another
still captures as it looks. -List prints every visible window with its
process name, size, and title: run it first when you are unsure of the
name. Never take a full-screen capture: it shows the user's other windows.
A menu, flyout, or tooltip that opens outside the window is not in the capture. Describe it in the index from the computer use screenshot and mark the section "no file". In a browser, use the tool's screenshot with a file path.
4. Walk the app in this order
- Main window, default page.
- Every navigation item (the left pane or sidebar), top to bottom. Expand collapsed groups.
- Every view mode (list, board, calendar), sort menu, filter menu.
- One item opened in detail: pane, fields, sub-items.
- Creation flows: add button, quick-add window (global shortcut), inline editor. Empty and half filled in, then Esc.
- Context menus (right-click, or Shift+F10) on an item, a section, a navigation entry.
- The menu bar or command bar: every menu once, and the "See more" (...) overflow. Press Alt to show access keys. Record every item, shortcut, and access key.
- Settings: every page.
- The tray icon in the notification area: its flyout and its right-click menu, if the app has one.
- Empty states.
- Dark mode: the two or three main pages again. Use the app's own theme setting. If it only follows Windows, ask the user to switch Settings > Personalization > Colors > Choose your mode.
Name files NN-page-or-state.png in capture order. The walk is
exhaustive when every navigation item and settings page has a file, and
every menu has an index section.
Lessons from real walkthroughs
- Many "Windows apps" are Electron. Their menu bars are nearly empty or hidden, and the actions hide in hover toolbars, right-click menus, and command palettes (often Ctrl+K or Ctrl+Shift+P). Hover rows and search the palettes (typing in a palette search is safe; do not press Enter).
- Store apps such as Calculator draw inside
ApplicationFrameHost, so their process name does not match. Match them by window title;-Listshows it. - A minimized window cannot be captured. Restore it first. Some windows (video players, protected content, some games) render black or blank: the script warns when the capture is one flat color.
- Preview fields test behavior without saving. A "Type a date" box in a to-do app shows how it parses a phrase before anything is stored. 30 phrases there corrected 3 answers from the docs.
- Note any change in the user's data during the walk, even if you are not sure you caused it, and tell them.
5. Write docs/screens/INDEX.md
One section per screenshot:
## 03-upcoming.png
Page: Upcoming. Reached from: navigation pane > Upcoming.
Layout: header with month picker and "Today" button; week strip; tasks
grouped by day, each day a section with a heading and an add button.
Elements: task row (checkbox, title, description line, date, project tag,
priority flag color).
Actions: complete, open, drag to another day, add to a day.
Shortcuts seen: none.End with three lists:
- Navigation map: navigation items and where each leads.
- Command map: every menu and command bar item, its keyboard shortcut, and its access key.
- Data model guess: the nouns and fields you saw.
Describe structure and behavior. Take no text, logo, or icon from the app:
rebuilding for personal use is fine, copying a brand is not. Keep the PNGs
out of public repos (add docs/screens/*.png to .gitignore).
6. Turn it into stories
Call the Skill tool with user-stories. Every screen in the index ends up
linked from at least one story, or listed under ## Dropped.
Done when INDEX.md has a section per PNG and per "no file" screen, the
three maps exist, and docs/STORIES.md covers every screen.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

