design-process-methods
Design process methodology: NNG's six phases, Double Diamond, Google Design Sprint, IDEO, Lean UX, and the bridge from AI-assisted building to evidence-based UX. Use when framing how the work should run, planning a sprint, or grounding a build in Empathize, Define, Ideate, Prototype, Test.
- License 许可
- MIT
- 最近更新
- 2026-11-10
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- design-process-methods
- description
- Design process methodology: NNG's six phases, Double Diamond, Google Design Sprint, IDEO, Lean UX, and the bridge from AI-assisted building to evidence-based UX. Use when framing how the work should run, planning a sprint, or grounding a build in Empathize, Define, Ideate, Prototype, Test.
- allowed-tools
- Read, Grep, Glob
02 资源文件 4
- 参考文档references/academic-foundations.md3.7 KB
- 参考文档references/company-ux-methodologies.md39.7 KB
- 参考文档references/nng-six-phases.md11.8 KB
- 参考文档references/vibe-coder-ux-bridge.md32.0 KB
03 SKILL.md 正文
UX Process Workflow — From Process to Product
Purpose
Ground every AI-assisted build in established UX process. Prompting can produce polished UI quickly; process decides whether it solves the right problem. The vibe-coding shortcut is:
Traditional: Empathize → Define → Ideate → Prototype → Test → Implement
Vibe coding: Prompt → Ship → Hope
The result is UI that solves the wrong problem, looks like every other AI-generated interface, and erodes the builder's own design judgment (arXiv:2509.10652).
The process in three lines
BEFORE you prompt: "Who is this for, and what problem does it solve?"
WHILE it generates: "Does this match the user's reality, or the AI's patterns?"
AFTER it generates: Run /audit + /a11y. Fix severity 3-4. Ship.
Core rules
- Use NNG's six phases in three stages. Understand (Empathize, Define) → Explore (Ideate, Prototype) → Materialize (Test, Implement). Design Ops is iterative, so loop back as needed. NNG provides the thinking; Design Ops commands provide the doing.
- Empathize before prompting. Name a specific user, the task (verb + noun), their current workaround, what frustrates them, and their environment (device, context, accessibility needs).
- Define in one sentence: "Users who [X] need [Y] because [Z]." If you can't fill it in, go back to Empathize. Add measurable success criteria.
- Diverge before converging. List three approaches before generating one. The first idea is rarely the best.
- Prototypes are questions. Generate 2-3 variations and evaluate them; don't treat the first output as production code.
- Testing is learning, not validation. Heuristic review, cognitive walkthrough, accessibility audit, and 5-user think-aloud tests find problems; they don't prove the design works.
- Watch for the three risks of skipping process: pseudo-productivity (more screens, fewer solved problems), homogenization (every AI UI looks the same), and skill erosion (losing the ability to critique).
- Check every generated screen for the persona, the information hierarchy, and all seven states: empty, loading, populated, error, partial, offline, updating.
Choosing a methodology
| Context | Methodology |
|---|---|
| Solo developer, vibe coding | Lean UX + Vercel/Figma design-engineer workflow |
| Small team (2-5) | Google Design Sprint (compressed) or Double Diamond |
| Large organization | IBM Enterprise Design Thinking |
| Accessibility-first | Microsoft Inclusive Design |
| Complex, ambiguous problem | IDEO Human-Centered Design |
| Continuous iteration | Spotify squad model + Lean UX |
The Builder's Cheat Sheet
I have 2 minutes
Run /brief on whatever you're about to build. Just the problem statement. That's it.
I have 10 minutes
/design-ops → /brief → then build with the constraint stack pasted into your prompt.
I have 30 minutes
/brief → /style → /tokens → then build. You now have a defined problem, visual direction, and token system.
I have 2 hours
Full Phase 0→5 journey. /design-ops → /brief → /style → /style → /tokens → /screen → /component → /audit → /a11y → /roast. This is what real design teams do.
I have a real product to ship
Follow the full 22-step journey via /design-ops. Run every audit. Loop /roast → /remix → /roast until all scores are 7+. This is how Stripe, Linear, and Notion work.
The one non-negotiable
Whatever your time budget: run /audit + /a11y before shipping. 5 minutes. Zero excuses.
Routing
| Question | Read |
|---|---|
| Each NNG phase: goals, methods, terminal workflow, command mapping; full process-to-command map | references/nng-six-phases.md |
| IDEO, Double Diamond, Design Sprint, IBM, Lean UX, Spotify, Microsoft, Vercel/Figma in depth; summary table; decision matrix | references/company-ux-methodologies.md |
| Before/while/after checklists, 15-minute UX check, anti-patterns, process recipes, maturity, constraint stack | references/vibe-coder-ux-bridge.md |
| ISO 9241-210, Norman's seven principles, cognitive laws | references/academic-foundations.md |
Cross-References
nng-ux-heuristics— The 10 heuristics used in the Test phaseux-research-methods— Research methodologies for the Empathize phasecognitive-psychology-ux— Cognitive laws that justify process rigoragentic-ai-generative-ux— AI-assisted design patterns and risksux-ethics-content-strategy— Ethical frameworks for the Define phaseux-metrics-measurement— Measurement frameworks for success criteriadesign-systems-architecture— System thinking for the Implement phasedesign-critique-case-studies— Case studies showing process impact (Stripe, Linear: process-heavy; Sonos 2024, Healthcare.gov: process-light failures)
04 同仓库更多 Skills
no-description
暂无简介
good-skill
Fixture skill that satisfies every rule.
broken-link
Links to a reference file that does not exist.
Bad_Name
Name violates the kebab-case rule and does not match the directory.
skill-authoring
Write, review, and validate portable agent skills in SKILL.md format. Use when creating a new skill, editing an existing one, or asked whether a skill follows the Agent Skills spec.
weigh-bags
Record scale readings for each bag — empty (tare), outbound, current, final — and reconcile them against the itemised inventory to find unaccounted weight. Use when the user gives a bag weight, says how heavy something is, weighs their luggage, or asks how close to the limit they are. Appends to bags.yaml and runs scripts/reconcile.py.