agentleFS
Eid0lonEid0lon/zero-slop/skills/perfect-design/SKILL.md

perfect-design

Skill13 starsChanged 6 months agoApache-2.0View on GitHub

Premium UI design engineering skill for creating, redesigning, polishing, and judging product-specific interfaces with senior-level craft. Use when Codex works on frontend design, UX/UI, design systems, dashboards, SaaS apps, landing pages, commerce, portfolios, visual polish, interaction quality, or wants output in the spirit of Vercel, Linear, Raycast, Awwwards, and high-end product teams. Composes with no-slop as the anti-generic gate.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 1

    1. lcsxu/zero-slop★ 13100% same

    Edited · 0

    None found.

    Wrote their own version · 1

    1. motionharvest/agent-skills★ 2same 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 repository4 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 8 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    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:

    Always write or infer a Design Contract before generating or rewriting UI.

    line 67
  2. Very specific:

    Always identify the user, job, domain expectation, design thesis, proof/data/media, and primary workflow.

    line 68
  3. Distinctive:

    Always inspect existing tokens, components, layouts, content, and states before editing.

    line 69
  4. Distinctive:

    Do not stop at removing cliches.

    line 71
  5. Distinctive:

    Do not use "premium", "clean", "modern", "beautiful", or "sleek" as design direction.

    line 72
  6. Distinctive:

    Do not make a landing page unless the user asked for marketing.

    line 73
  7. Distinctive:

    Do not hide the product, object, place, app, game, state, or primary task in the first viewport.

    line 74
  8. Distinctive:

    Do not ship card soup, decorative gradients, stock-like blobs, fake proof, vague CTAs, or motion without purpose.

    line 75
  9. Distinctive:

    Use proven primitives for difficult interaction patterns when the stack provides them.

    line 78

≈ “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

Perfect Design

Perfect Design is the premium design layer. no-slop removes generic AI residue; Perfect Design creates a coherent, product-specific interface that feels designed by a senior design engineer.

Command Interface

Use references/command-interface.md.

Primary commands:

perfect-design --contract [brief-or-target]
perfect-design --create [brief-or-target]
perfect-design --redesign [target]
perfect-design --polish [target]
perfect-design --judge [target]
perfect-design --verify [target]
perfect-design -e --polish [target]

Short flags:

  • -c, --contract: write the design contract before code.
  • --create: build a new UI surface from a brief.
  • -r, --redesign: rebuild direction, layout, tokens, copy, and interaction.
  • -p, --polish: refine an existing surface without changing the product model.
  • -j, --judge: run the brutal premium judge panel.
  • --verify: prove the result in code and browser when possible.
  • -e, --economy: disable live subagent judges only when the user explicitly accepts a lower-cost review.

Default mode:

  • Existing UI target: --polish.
  • Brief without target: --contract.

First Action

Load steps/step-00-command-router.md.

Step 00 parses command flags, discovers project shape, loads the right references, and routes to contract, create, redesign, polish, judge, or verify.

Core References

Load only what the task needs:

  • references/command-interface.md: grammar, precedence, output contracts.
  • references/dials-and-presets.md: premium craft dials and archetype presets.
  • references/research-synthesis.md: distilled research from Vercel, Linear, Raycast, Awwwards, WCAG, Baymard, Polaris, Radix, web.dev, Resend, Supabase, and related sources.
  • references/design-contract.md: pre-code contract structure and rejection rules.
  • references/product-archetypes.md: domain-specific design expectations.
  • references/composition-playbook.md: layout, hierarchy, typography, color, tokens, imagery, and visual rhythm.
  • references/interaction-accessibility.md: keyboard, focus, states, motion, performance, and responsive stability.
  • references/no-slop-integration.md: how to combine Perfect Design with no-slop.
  • references/legacy-boundary.md: proof that old-perfect-design is intentionally outside the editable scope.
  • references/judge-system.md: mandatory brutal judge protocol and gate formula.
  • references/premium-quality-rubric.md: scoring rubric and ship gates.
  • references/self-coaching.md: the internal checklist to use when stuck.
  • references/verification-playbook.md: build/browser/accessibility/visual proof.

Non-Negotiables

  • Always write or infer a Design Contract before generating or rewriting UI.
  • Always identify the user, job, domain expectation, design thesis, proof/data/media, and primary workflow.
  • Always inspect existing tokens, components, layouts, content, and states before editing.
  • Compose with no-slop whenever available. Run it before and after material UI work, or mark NO_SLOP_UNAVAILABLE.
  • Do not stop at removing cliches. Replace them with clear product decisions.
  • Do not use "premium", "clean", "modern", "beautiful", or "sleek" as design direction. Translate style words into concrete type, spacing, density, palette, material, copy, motion, and state decisions.
  • Do not make a landing page unless the user asked for marketing. Build the actual usable surface first.
  • Do not hide the product, object, place, app, game, state, or primary task in the first viewport.
  • Do not ship card soup, decorative gradients, stock-like blobs, fake proof, vague CTAs, or motion without purpose.
  • For operational tools, prefer dense, scan-friendly, restrained interfaces. For expressive/editorial work, create a deliberate art direction with strong content rhythm.
  • Preserve behavior, accessibility, semantics, keyboard access, focus visibility, responsive stability, and performance.
  • Use proven primitives for difficult interaction patterns when the stack provides them.
  • Judge with subagents by default. If live judges are unavailable and economy mode is false, mark JUDGE_PANEL_UNAVAILABLE; never claim a full pass.
  • Iterate until the premium gate passes. If blocked, state exact blockers.

Required State

Maintain this state throughout the workflow:

command:
target:
mode:
economy_mode:
preset:
dials:
project_type:
user:
primary_job:
design_thesis:
design_contract:
no_slop_before:
premium_score_before:
files_changed:
judge_panel:
judge_gate:
verification:
no_slop_after:
premium_score_after:
remaining_risk:

Local CLI Helper

For deterministic prechecks:

python perfect-design/scripts/perfect_design_cli.py --contract "brief text"
python perfect-design/scripts/perfect_design_cli.py --audit <target>
python perfect-design/scripts/perfect_design_cli.py --preset dashboard --dial CRAFT_STRICTNESS=10 --audit <target>

The CLI is a precheck, not a replacement for the Codex workflow, live browser review, or subagent judging.

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.