agentleFS
emanueledenaroemanueledenaro/claude-skills/screen-map/SKILL.md

screen-map

Skill0 starsChanged 36 days agoMITView on GitHub

Build and keep a product screen map that every builder follows as the source of truth: one page with every screen per platform, real copy, a status per screen (exists, built to test, proposal), arrows for the flow, simulations of real users, a style guide, and a project skill generated from it. Use when asked for a map of the screens, a UI flow, a source of truth for app, menu, site or onboarding, to turn a map into a skill, or to keep design and code aligned, also in Italian (mappa delle schermate, flusso, schermate dell'app, fonte di verità, trasforma la mappa in una skill), in any project.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 9 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    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
  1. Distinctive:

    Lanes: follow the phases of use: install, first action, while it works, settings, and so on.

    line 10
  2. Distinctive:

    Add one lane for the style guide, one for the website, and one for each simulation.

    line 12
  3. Distinctive:

    Real data first: Before drawing, check copy, colours and behaviour against the code and against photos of the real app.

    line 28
  4. Distinctive:

    The map changes before the code: When a decision is taken, update the map first.

    line 29
  5. Distinctive:

    A photo of the real device beats the drawing: When a test on a real device shows a difference, ask what the person wants.

    line 30
  6. Distinctive:

    Values come from the real thing: When the person says "make it like X", sample X from the photo or the code.

    line 31
  7. Distinctive:

    Check every frame against these before it goes on the map, then against the project's own design skill if it has one (for example a taste …

    line 44
  8. Very specific:

    If the project has a skill made from the map, generate it again (reference.md section 4.

    line 61
  9. Distinctive:

    Before a map's content goes anywhere other people can see, remove prices that are not yet announced, PR numbers, code internals, supplier …

    line 70
  10. Distinctive:

    Every screen has a status, real copy and notes, on every platform where it exists, and passes the design checklist.

    line 74

≈ “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

A screen map is the reference every builder follows: the app, its menus, the website, and anything else the person sees. It is not a mood board. Each screen on it says what is true today, in the product's own words. The entry format, the page layout, the handoff message and the generator for the project skill are in reference.md. Read it when you create a map, add screens, or turn a map into a skill.

1. What the map holds

  • Lanes follow the phases of use: install, first action, while it works, settings, and so on.
    • Give each platform its own set of lanes when the platforms differ (macOS, Windows, web, mobile).
    • Add one lane for the style guide, one for the website, and one for each simulation.
  • One frame per screen, drawn inside the real system frame (menu bar, taskbar, browser, phone). Each frame has a number (3.4), a stable id (mac-panel, win-panel) and a short title.
  • One status per frame, never mixed:

    • exists means it is on the main branch.
    • built, to test means it is in a pull request and has not yet been tried on a real device.
    • proposal means it is designed but not built.

    A proposal is never shown as built.

  • Real copy: the exact words of the interface, in the interface's language.
  • Arrows show the flow. Each one is labelled with the action that moves the person ("Stop", "Esc", "Buy").
  • Details for each screen: what it does, what to know, what is still open, and which plan or tier it belongs to if the product has tiers.
  • Simulations: one real kind of user doing one real task from start to finish. Features come out of simulations, not out of generic feature lists.
  • Style guide: colour tokens with their roles, type, corners per platform, and the rules that apply on every screen.

2. Rules

  • Real data first. Before drawing, check copy, colours and behaviour against the code and against photos of the real app. Mark anything you have not checked as an assumption.
  • The map changes before the code. When a decision is taken, update the map first. Then tell builders exactly what changed (values, copy, behaviour), not only "see the map".
  • A photo of the real device beats the drawing. When a test on a real device shows a difference, ask what the person wants. Then fix the map and pass the fix on.
  • Values come from the real thing. When the person says "make it like X", sample X from the photo or the code. Do not use a guessed colour, such as white or black.
  • Decisions are dated in the screen notes ("Decided 2026-11-04: …"). Remove the note a decision replaces; do not leave both.
  • Open points are listed, not solved silently. A question that changes the goal goes to the person once, with a recommendation.
  • One visual language. A signal means the same thing on every screen, and a colour never carries a state on its own.
  • One page that also works on a phone, with:
    • pan and zoom;
    • a sidebar listing every screen;
    • a detail panel;
    • a link to each screen (#id);
    • a platform switch when the screens differ by platform.

3. Design checklist

Check every frame against these before it goes on the map, then against the project's own design skill if it has one (for example a taste or brand skill):

  1. Visual hierarchy: one element catches the eye first, set apart by size, weight and colour.
  2. Contrast: light and dark, large and small, clearly apart. Every text tier reaches 4.5:1 on its surface.
  3. Alignment and grid: one left edge per surface, one spacing base (4px or 8px), the same padding on surfaces of the same kind.
  4. White space: groups are separated by space before lines or boxes, so the screen breathes and the main subject stands out.
  5. Balance: visual weight is spread so that the screen is steady, with no side empty and no side crowded.
  6. Typography: one text face, a few sizes, nothing under 11px, tabular figures for numbers, readable line length.
  7. Consistency: the same button, colour, radius and spacing everywhere, and one meaning per signal.

Native interface parts follow the platform's own materials and controls (system glass, menus, title bars). Mark that as a platform choice, not a style of your own.

4. After every change

  1. Render the frames you changed and look at them before you publish.
  2. Publish the new version at the same link, so links that were already shared keep working.
  3. Tell whoever tracks status what changed. Send builders the spec of the change.
  4. If the project has a skill made from the map, generate it again (reference.md section 4; this skill's section 5), so the text and the drawing never disagree.

5. Turning a map into a skill

A map becomes a project skill. It lives in the project's own repository, for example .claude/skills/<product>-map/, and never in a shared or public skills repository. Product facts stay with the product. The skill has:

  • a SKILL.md with the global rules, the colour tokens, the status legend, and how to work with the map;
  • one reference file per platform or lane group, generated from the map's data, with the exact copy for each screen.

Before a map's content goes anywhere other people can see, remove prices that are not yet announced, PR numbers, code internals, supplier names and real people's names.

Done when

  • Every screen has a status, real copy and notes, on every platform where it exists, and passes the design checklist.
  • Nothing on the map contradicts the latest decision or the latest test on a real device.
  • Builders have been told what changed, and any skill made from the map has been generated again.

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.