agentleFS
cheetah-tradecheetah-trade/claude-code-team-kit/skills/layout-and-hierarchy/SKILL.md

layout-and-hierarchy

Skill0 starsChanged 29 days agoMITView on GitHub

Lay out a screen, block or card so a person reads it in seconds: one headline per block, rows and cards over prose, one-line titles with state on the right, scannable density, empty states, first action first. Use when designing or reviewing any screen, especially one that shows generated text such as results, summaries or reports.

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, 10 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use when you design or review any screen, block or card, and above all when the content is generated text (results, summaries, notes, …

    line 8
  2. Distinctive:

    One headline per block: Every block opens with one line that says its outcome in plain words ("Import finished.

    line 16
  3. Distinctive:

    Lists and cards over prose: When content has items (links, files, criteria, steps, decisions), each item is a row or a card, never a …

    line 19
  4. Distinctive:

    An import result: Before: "The import completed with 1,204 rows processed, of which 3 had invalid postcodes (rows 18, 77 and 940) and 12 …

    line 36
  5. Distinctive:

    An orders dashboard: Before: a 12-column table where most cells are "—" and every row has the same "Paid" badge.

    line 41
  6. Distinctive:

    After: "Needs action" on top (failed payments, refund requests) as one-line rows with the action on the right.

    line 42
  7. Distinctive:

    Write the headline first: what is true and what the user must do, in one line.

    line 47
  8. Distinctive:

    Write the empty state of every group and of the whole block.

    line 51
  9. Distinctive:

    Read your own screenshot for five seconds as the user.

    line 54
  10. Distinctive:

    Every block has a headline.

    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

Layout and hierarchy

Use when you design or review any screen, block or card, and above all when the content is generated text (results, summaries, notes, reports). Pair it with ui-review-checklist and component-states.

People read a screen in seconds, between other work. A screen succeeds when they know in one glance what is done, what needs them and what they press first. A wall of text fails even when every word in it is true.

The six rules

  1. One headline per block. Every block opens with one line that says its outcome in plain words ("Import finished. 3 rows need your review", "2 invoices overdue"). Everything under it supports that line. A block without a headline, or with two lines competing for it, is a defect.
  2. Lists and cards over prose. When content has items (links, files, criteria, steps, decisions), each item is a row or a card, never a sentence in a paragraph. Prose is fine for one or two sentences of explanation, never for something the user opens, checks or acts on.
  3. One-line titles, state on the right. A row is: kind (icon or label) · title cut to one line with an ellipsis (full text in a tooltip and in the detail view) · state on the right edge, short and always in the same place, coloured only when it needs the user (warning) or is good news (success). The same state uses the same word everywhere ("shipped", not "Shipped" here, "sent" there, "dispatched" elsewhere).
  4. Scannable density. Rows of one height, aligned columns, groups with a small muted label above. No dead space: a mostly empty column or block gives its width back; a block with nothing in it is hidden or shows a one-line empty state. No crowding either: a row carries a title, one meta and one state, not five chips.
  5. Empty states say why and what next. An empty list is a line or a small card: what would be here, why it is empty, the first action. A blank area, a lone heading or "—" in every cell is a defect.
  6. What the user does first is first. Order blocks by what it costs to miss them: needs you → failed or stuck → what changed → reference. The primary action is one button named for its result; others are quieter.

Before and after

An import result. Before: "The import completed with 1,204 rows processed, of which 3 had invalid postcodes (rows 18, 77 and 940) and 12 were duplicates that were skipped; you can download the log here." After: the headline "Import finished · 3 rows need your review"; under it three rows (row number, customer name, "invalid postcode", Fix), then a quiet line "12 duplicates skipped · Download log". Same facts, read in two seconds.

An orders dashboard. Before: a 12-column table where most cells are "—" and every row has the same "Paid" badge. After: "Needs action" on top (failed payments, refund requests) as one-line rows with the action on the right; below, recent orders with four columns; quiet days folded into one line ("4 days with no orders").

How to design a block (in this order)

  1. Write the headline first: what is true and what the user must do, in one line.
  2. List the items the content holds; give each a row or card with title · meta · state; fix the state word set.
  3. Group them under small labels; order groups by rule 6; fold reference material (old versions, quiet items, details) behind a disclosure with a count ("Show 8 earlier versions").
  4. Write the empty state of every group and of the whole block.
  5. Draw it at desktop and phone width, light and dark; on the phone the state moves under the title rather than squeezing it.
  6. Read your own screenshot for five seconds as the user. If you cannot say what needs them, start again.

Review: structure defects to look for

DefectWhat it looks likeFix
Text wallitems, links or decisions inside a paragraphone row per item, headline above
No hierarchyheadline, labels and body in the same size and weightstronger headline, small muted labels, one body size
Mixed statesthe same state in different words, colours or positionsone word set, always on the right
Cut labeltext clipped mid-word with no ellipsis, or truncated until it loses meaningellipsis with a tooltip, or wrap; shorten the copy
Dead spacewide empty columns, a block mostly blank, items squeezed beside itgive width back, collapse empty groups
Missing empty stateblank area, lone heading, "—" everywhereone line: why and the first action
Buried actionwhat the user must do sits in the middle or at the endfirst block, one primary button
Chip noisebadges that do not change what the user doesremove them

Name the exact element by its visible text ("the row 'Order #1042'", "the Recent orders label"), not the class of problem in general. A screen with no defect is a valid finding: say it is clean, do not pad.

Done when

Every block has a headline; every actionable thing is a row or card; states use one word set on the right; nothing is clipped without an ellipsis; no block is mostly empty; every list has its empty state; light, dark, desktop and phone width checked from screenshots.

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.