agentleFS
fluxnative-appfluxnative-app/fluxnative-claude/skills/design-references/SKILL.md

design-references

Skill0 starsChanged 10 days agoMITView on GitHub

Use when designing, restyling or recreating screens in a FluxNative app, or when the user names an app or flow to emulate: search FluxNative's catalog of real app journeys and screens, then adapt structure and density to the app's own theme, never copying another product.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. anis-dr/skills★ 4same 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 repository9 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 5 are distinctive, and 2 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.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write a query as one sentence about one subject: what happens in the journey, or what appears on the screen and how it is composed.

    line 26
  2. Very specific:

    Find the pattern: Search the journey with search_flows, and each screen kind with search_screens.

    line 37
  3. Distinctive:

    Stop: when someone holding both images would match every zone and the anchor without hesitating.

    line 47
  4. Distinctive:

    Never copy another product's brand, copy, imagery, icons or logo into the app.

    line 59
  5. Very specific:

    Never add a second icon family, or hard-code colors and sizes outside src/theme.

    line 60
  6. Distinctive:

    Never fetch screens you are not about to build: each one costs vision tokens.

    line 61
  7. Distinctive:

    Never draw standard platform chrome by hand when the reference shows a standard tab bar or header.

    line 62

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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 real app references

When to use

  • A new screen or flow in a FluxNative app, where real products show how the journey usually goes (onboarding, checkout, settings, a feed).
  • The user names an app to look like, or attaches a screenshot to match.
  • A screen looks generic and you need evidence of how good apps solve it.

The catalog is the same for every project, so its searches take no project. It holds mobile apps: use it for an app, not a landing page, which has its own template library (search_landing_templates, in the app-workflow skill). Reading and changing the app's own source takes its id as project (the app-workflow skill).

The catalog

FluxNative keeps a catalog of real mobile apps, broken into journeys (flows) and screens. Find references by the screen pattern or the journey you are building, not by the app: the catalog is thin per app and rich per pattern. Two searches reach it, each by a prose query or by the ids a result returned:

ToolReturnsCost
search_flowsJourneys, each with its full ordered list of screen idsText only
search_screensScreens as inline imagesA few hundred vision tokens per screen: about 260 on Claude for 299×648, about 760 for a 512-wide screen
  • Every flow and screen names its app: app_id, app_name and a one-line app_summary. Pass app (an app_id) to either search to keep to that one product, for example to pull more of one app's screens so a journey stays consistent; app alone lists that app's journeys or screens.
  • Write a query as one sentence about one subject: what happens in the journey, or what appears on the screen and how it is composed. Keyword lists, negations and style words rank badly.
  • Compose each screen from exemplars of its kind, across apps. An empty result is ordinary: describe the screen another way, or ask the other search.
  • A screen with an exemplar_id is FluxNative's own coded screen: start the feature file from it with copy_template (exemplar:<slug>), never by retyping it, then change its data, copy and brand.
  • A screen marked identity: "open" may also inform the design DNA's palette family, illustration and density, never a mark or logo.
  • load_skill holds the design procedures (design-direction, anti-slop-invariants, accessibility-pass, …).
  • Fetch one journey's screens (search_screens with the step ids), build that journey, then come back for the next. Never fetch a whole app's screens at once. search_screens with exclude_ids skips screens you have already seen.
  • Hosts that render MCP Apps show flows and screens in an interactive view, where the user can pick a screen or flow and hand its id back to you.

Procedure

  1. Read the app first. src/theme and @/components define what you can use, and the app's design DNA says who it is: .fluxnative/memory/design-dna.md, with design-screens.md, design-components.md and product-dna.md beside it when the app has them. Note the palette roles, type scale, radii and icon family. Identity comes only from these files, never from a reference.
  2. Find the pattern. Search the journey with search_flows, and each screen kind with search_screens. Pick two or three references that fit the app's audience and density; they may come from different apps.
  3. Write down what you will take from each screen:
    • the zones from top to bottom, each as a share of the height;
    • the anchor and the reading order;
    • the repeated units (row, card, chip) and their anatomy;
    • what stays fixed and what scrolls;
    • the states shown (selected, empty, error) and where each control leads.
  4. Map it onto the app's tokens. Colors land on theme roles, type sizes on the nearest type roles, spacing and radii on their scales. Change a role once in src/theme when the whole app should shift; never scatter literals.
  5. Build structure first, with the app's real copy and its kernel components, then compare in the live preview (the preview-verification skill) on the closest device.
  6. Fix differences in order: structure and proportions, then type, then color and surfaces, then detail (icons, crops, borders). Re-check after each tier, not after each fix.
  7. Stop when someone holding both images would match every zone and the anchor without hesitating. Note what differs on purpose.

What a reference gives you

ReferenceTakeLeave
A catalog journey or screenStructure, hierarchy, density, navigation, the order of stepsIts palette, brand, copy, imagery and exact geometry
A screenshot of the user's own productEverything visible: it is the specNothing; ask about what it does not show
A screenshot of someone else's appStructure and density, as with the catalogBrand, copy, imagery, logos

Never

  • Never copy another product's brand, copy, imagery, icons or logo into the app.
  • Never add a second icon family, or hard-code colors and sizes outside src/theme.
  • Never fetch screens you are not about to build: each one costs vision tokens.
  • Never draw standard platform chrome by hand when the reference shows a standard tab bar or header; use the platform's.

Report

  • The references you used (app and journey names) and what you took from each.
  • Token changes made in src/theme, if any.
  • Deliberate differences from the references, and anything not yet checked in the preview.

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.