agentleFS
daveebbelaardaveebbelaar/bama/skills/design/design-from-app/SKILL.md

design-from-app

Skill1.7k starsChanged 12 days agoMITView on GitHub

Rebuild the look of an app the user likes as a native SwiftUI design kit, by measuring its screenshots (sizes, spacing, type, colors, icons), building its parts, and comparing pixel by pixel until they match. Use when the user names an app whose look they want, shares screenshots of one, or says "make it look like X".

2 other projects use a version of this, 2 of them edited.

Adopted by

Steady: 0 new projects took it in the last 4 weeks, none the 4 weeks before.thin data

Week of Sep 13: 0 new projectsWeek of Sep 20: 0 new projectsWeek of Sep 27: 0 new projectsWeek of Oct 4: 0 new projectsWeek of Oct 11: 0 new projectsWeek of Oct 18: 0 new projectsWeek of Oct 25: 0 new projectsWeek of Nov 1: 0 new projects
Sep 2026Oct 2026Nov 2026
New projects taking a version of this file each week, over the 8 weeks our latest run read: only 0 in all, so this is a hint, not a trend.

Most recently taken up by

Who's talking about daveebbelaar/bama

The biggest 5 of 42 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 42

Talked about on GitHub, YouTube and the web (42 times).

On GitHub · 20

15 more on GitHub

On YouTube · 13

8 more on YouTube

On the web · 9

4 more on the web

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. About 0 are common, 5 are distinctive, and 10 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 10

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Look at an app, measure it, and rebuild its design system in SwiftUI.

    line 9
  2. Very specific:

    Run them with swift "${CLAUDE_SKILL_DIR}/scripts/measure.swift" ....

    line 21
  3. Distinctive:

    All coordinates are points (pixels / 2 for a Retina capture).

    line 22
  4. Very specific:

    Ask for screenshots, or take them with the clone skill.

    line 26
  5. Very specific:

    Check: info prints the window size in points, and every screenshot uses the same scale.

    line 33
  6. Very specific:

    Look at each screenshot and name every distinct part: window chrome, sidebar row, section header, primary action, list row, checkbox, date …

    line 38
  7. Distinctive:

    Check: every visible element on the main screenshot belongs to a part.

    line 43
  8. Very specific:

    Check: every token and every part size in design/KIT.md has a number from a command.

    line 60
  9. Distinctive:

    When the letters differ (a, g, t, digits), name the font and pick it or an open alternative with a license that allows bundling.

    line 66
  10. Very specific:

    Check: design/KIT.md names the font, the icon source, and the icon size and weight.

    line 74
  11. Very specific:

    Call the Skill tool with design-kit and build the tokens, the parts, the window chrome, and the gallery from design/KIT.md.

    line 79
  12. Distinctive:

    Check: a snapshot of that screen has the same size in pixels as the reference.

    line 84
  13. Distinctive:

    Check: under 2% of pixels differ on the whole window, and every row and left edge is within 1.5 pt of the reference.

    line 99
  14. Very specific:

    Before the app ships to anyone else, replace what belongs to the other company: its name, logo, illustrations, and any font or icon it owns.

    line 104
  15. Very specific:

    Keep the structure, the spacing, and the feel.

    line 106

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Design from app

Look at an app, measure it, and rebuild its design system in SwiftUI. The result is a design kit (design-kit skill) with the reference app's numbers, and one reference screen that matches the screenshot.

Guessing gets close. Measuring gets it right. Two scripts do the measuring, next to this skill:

  • scripts/measure.swift: sizes, positions, text heights, and colors from a screenshot.
  • scripts/compare.swift: a side-by-side, a diff, and the share of pixels that differ.

Run them with swift "${CLAUDE_SKILL_DIR}/scripts/measure.swift" .... All coordinates are points (pixels / 2 for a Retina capture).

1. Get the screenshots

Ask for screenshots, or take them with the clone skill. You need:

  • The window only, at 2x, no desktop. Save them in design/reference/.
  • One screen with the most parts on it (a sidebar, a list, a toolbar), plus close-ups of menus, dialogs, and dark mode if they matter.
  • The window size: measure.swift info shot.png.

Check: info prints the window size in points, and every screenshot uses the same scale.

2. List the parts

Look at each screenshot and name every distinct part: window chrome, sidebar row, section header, primary action, list row, checkbox, date chip, icon button, divider. Start design/KIT.md from the design-kit skill's kit-spec.md with one section per part.

Check: every visible element on the main screenshot belongs to a part.

3. Measure

Fill design/KIT.md with numbers, not estimates. The recipes and the traps are in measuring.md. In short:

To findCommand
Sidebar width, paneshline shot.png <y in an empty area>
Row positions and pitchrows shot.png <x0> <x1> [y0 y1]
Left edges, gapscols shot.png <y0> <y1> [x0 x1]
Selection pill, dividersvline and hline through it, --tolerance 4
Text and icon colorsink shot.png x y w h
Surface colorspalette shot.png x y w h
A close lookcrop shot.png x y w h out.png 4

Check: every token and every part size in design/KIT.md has a number from a command.

4. Find the font and the icons

  • Most apps on the Mac use the system font, even web apps. Start there. When the letters differ (a, g, t, digits), name the font and pick it or an open alternative with a license that allows bundling.
  • Text sizes come from step 6: build, measure the same word in both images, scale.
  • Icons: note the size, stroke weight, corners, and whether selected rows switch to a filled form. Match with SF Symbols at a light weight first; draw the ones that do not match as shapes (design-kit parts.md, Icons).

Check: design/KIT.md names the font, the icon source, and the icon size and weight.

5. Build the kit and one screen

Call the Skill tool with design-kit and build the tokens, the parts, the window chrome, and the gallery from design/KIT.md. Then build the main screenshot's screen from kit parts only, with the same data and the same window size.

Check: a snapshot of that screen has the same size in pixels as the reference.

6. Compare until it matches

swift "${CLAUDE_SKILL_DIR}/scripts/compare.swift" design/reference/main.png /tmp/<App>-shots/<page>-light.png /tmp/<App>-compare
swift "${CLAUDE_SKILL_DIR}/scripts/compare.swift" ... /tmp/<App>-compare-sidebar 0 0 220 300

Open side-by-side.png and diff.png with the Read tool. Then measure the same rows, edges, and words in both images, and fix the tokens, not the views. Order: window size, text sizes, columns and left edges, row heights, colors, icons.

Check: under 2% of pixels differ on the whole window, and every row and left edge is within 1.5 pt of the reference.

7. Make it yours

Before the app ships to anyone else, replace what belongs to the other company: its name, logo, illustrations, and any font or icon it owns. Keep the structure, the spacing, and the feel. Write the replacements in design/KIT.md.

Done when design/KIT.md has a measured number for every token and part, the kit's gallery shows every part, the reference screen passes step 6, and step 7 lists what was replaced.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.