Skip to content
Back to search

design-craft

Agent skills for designed, not generated, frontends. UI/UX, design systems and motion for Claude Code and Codex.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/iitian001--design-craft/badge.svg)](/listings/iitian001--design-craft)

HTML

<a href="/listings/iitian001--design-craft"><img src="/api/listings/iitian001--design-craft/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

**Claude Code and Codex, one command:**

git clone https://github.com/Iitian001/design-craft.git
cd design-craft
./install.ps1          # Windows
git clone https://github.com/Iitian001/design-craft.git && cd design-craft && ./install.sh   # macOS / Linux

The installer copies every skill in `skills/` to `~/.claude/skills/` and, if it exists, `~/.codex/skills/`. Re-run it after `git pull` to update.

**Manual:** copy `skills/design-craft` into `~/.claude/skills/` (Claude Code) or `~/.codex/skills/` (Codex).

**Use it:** type `/design-craft` in Claude Code, or `$design-craft` in Codex. Agents also pick it up automatically for UI, UX and motion work.

**Optional, for verification:** `pip install playwright` (it uses your installed Chrome or Edge).

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 104 lines
  1. # Design Craft
  2. **Agent skills for building frontends that are designed, not just generated.**
  3. Works with **Claude Code** and **Codex**.
  4. `design-craft` is a working method an AI coding agent follows when it designs, builds, reviews or fixes
  5. UI: websites, landing pages, product screens, components, design systems, animation and scroll scenes.
  6. It's distilled from nine classic design books and sharpened by real builds, where every bug that slipped
  7. through became a rule.
  8. ```
  9. Frame → Find the signature → Explore cheaply → System first → Every state
  10. → Emotion, copy & honesty → Motion → Verify in a real browser → Hand off
  11. ```
  12. ## What it makes the agent do differently
  13. - **Frames the job before coding:** a job story, the ambition the brief actually asks for, and the
  14. visitor's emotional state on arrival.
  15. - **Finds a signature interaction in the subject** (a beer pours, a fabric drapes, a typeface flexes) and
  16. runs the *substitution test*: if the effect would work unchanged for any brand, it's not done.
  17. - **Builds tokens before components**, covers every state (focus, loading, empty, error, offline), and
  18. makes the newest async request win.
  19. - **Gives every animation a job**, with motion tokens, correct easing direction, interruption handling and a
  20. reduced-motion path. It also never makes people wait on choreography before they can act.
  21. - **Avoids engineering traps that broke real builds:** GSAP fighting CSS over a transform, scroll timelines
  22. that autoplay, snap labels past the timeline end, negative frame time, swappable assets that weren't preloaded.
  23. - **Verifies in a real browser** with an included Playwright walker: desktop, phone and reduced motion, a real
  24. pointer drag, console errors and sideways overflow.
  25. - **Stays honest:** labels AI imagery and demo data, makes copy match the images, and applies category rules
  26. (alcohol, finance, health).
  27. ## Install
  28. **Claude Code and Codex, one command:**
  29. ```powershell
  30. git clone https://github.com/Iitian001/design-craft.git
  31. cd design-craft
  32. ./install.ps1 # Windows
  33. ```
  34. ```bash
  35. git clone https://github.com/Iitian001/design-craft.git && cd design-craft && ./install.sh # macOS / Linux
  36. ```
  37. The installer copies every skill in `skills/` to `~/.claude/skills/` and, if it exists, `~/.codex/skills/`.
  38. Re-run it after `git pull` to update.
  39. **Manual:** copy `skills/design-craft` into `~/.claude/skills/` (Claude Code) or `~/.codex/skills/` (Codex).
  40. **Use it:** type `/design-craft` in Claude Code, or `$design-craft` in Codex. Agents also pick it up
  41. automatically for UI, UX and motion work.
  42. **Optional, for verification:** `pip install playwright` (it uses your installed Chrome or Edge).
  43. ## What's inside
  44. ```
  45. skills/design-craft/
  46. ├── SKILL.md the method: 8 steps, loaded every use (~2.5k tokens)
  47. ├── agents/openai.yaml Codex display metadata
  48. ├── references/ loaded only when a step needs depth
  49. │ ├── motion-engineering.md GSAP / ScrollTrigger / Lenis / canvas / gesture traps
  50. │ ├── verification.md what proves what, and environment traps
  51. │ ├── imagery.md generating and preparing images, product cut-outs
  52. │ ├── animation.md the 7 motion principles, tokens, easing, stagger maths
  53. │ ├── emotion.md persona, peak-end, trust, delight, error and outage copy
  54. │ ├── design-systems.md tokens, components, governance, versioning
  55. │ ├── product-design.md research, story-first, sketching, prototyping, testing
  56. │ ├── design-thinking.md empathise → define → ideate → prototype → test
  57. │ ├── design-engineering.md prototype fidelity, the design↔code handshake
  58. │ └── process-and-teams.md critique, rationale and handoff templates, DesignOps
  59. └── scripts/walk.py Playwright walker for real-browser verification
  60. evals/briefs/ benchmark briefs used to measure each version
  61. ```
  62. ## How it improves
  63. Every change has to earn its place:
  64. 1. Run a benchmark brief from `evals/briefs/` with the current skill.
  65. 2. Verify the result with `scripts/walk.py` and review it against the checklist in `SKILL.md` step 8.
  66. 3. Turn each real failure into **one** rule at the right level: a fix in the script first, then a reference
  67. file, and `SKILL.md` only if every task needs it.
  68. 4. Record what changed and why in `CHANGELOG.md`.
  69. See [CONTRIBUTING.md](CONTRIBUTING.md) and [ROADMAP.md](ROADMAP.md). Backend and system-design skills are next.
  70. ## Sources
  71. The principles are paraphrased and synthesised from these books. No text is reproduced beyond short
  72. phrases. The books are excellent and worth reading in full.
  73. - *Designing for Emotion*, Aarron Walter
  74. - *Animation Handbook*, *Design Engineering Handbook*, *Collaboration Handbook*, *Design Leadership Handbook*,
  75. *Design Thinking Handbook*, *DesignOps Handbook*, *Design Systems Handbook* (DesignBetter / InVision)
  76. - *Principles of Product Design* (InVision)
  77. This project is independent and not affiliated with or endorsed by the authors or publishers.
  78. ## License
  79. [MIT](LICENSE)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.