UX Audit — All expertise rules

Layers: L1 ✅ · L2 desktop dark ✅ · L3 list → detail ✅.

Patterns in use

Pattern	Rating	Evidence
Breadcrumbs / Deep-linking	✅	Domain → all rules → rule has stable routes.
Overview + Detail	✅	Every row is a semantic Link.
Titled Sections	⚠️	Page “全部经验” and inner “经验” repeat hierarchy.
Dynamic Query	⚠️	Tier exists; service search/layer are not exposed.
List at scale	—	All rows render eagerly; no pagination/virtualization/search contract.
Failure + Retry	⚠️	Domain request is covered; lessons request is not.

Strengths / good cases

• ✅ 亮点 — Rows are semantic navigation. Runtime accessibility exposes P-01…P-13 as links.
• ✅ 亮点 — Hit ratio stays at the scan edge. Stable columns align identity, title and usage.

Experience gaps

1. 🔴 Lesson failure renders an empty body under a non-zero total. Lessons error/loading are discarded (RulesDetail/index.tsx:32-33, :83-87).
2. 🟠 Not designed beyond fixture scale. No search, layer/anchor filter, sort, paging or virtualization.
3. 🟡 Page and card restate the same title. Replace card shell with page toolbar + list.
4. 🟡 Filter state is local, resets, and is not deep-linkable. Move read state into URL/server query.

Skill feedback

• Validates Read §§1.1, 1.2 and URL-backed multi-dimensional list state.