UX Audit — Expertise rule detail

Layers: L1 ✅ · L2 desktop dark ✅ · L3 direct URL/list/breadcrumb ✅.

Patterns in use

Pattern	Rating	Evidence
Sequence Map / Breadcrumbs	✅	Expertise → domain → all rules → P-code.
Titled Sections	✅	Structured sections adapt to polarity.
Attached proof	⚠️	Hit examples are shown but not linked to source Topic.
Empty state	⚠️	No-examples is explicit but has no explanation/action.
Failure + Retry	✅	Rule fetch distinguishes loading/error/not-found and retries.

Strengths / good cases

• ✅ 亮点 — Error and deletion differ. Request error and resolved absence have separate states (LessonDetail/index.tsx:84-93).
• ✅ 亮点 — Structure follows polarity. Stored section keys define the reading structure (:112-123).
• ✅ 亮点 — Deep-linkable reading chain. List row, URL and breadcrumb preserve hierarchy.

Experience gaps

1. 🟠 Proof cannot be audited at source. Backend returns subjectId/subjectType, but UI prints a title only (:132-155). Add source Topic link/date/unavailable state.
2. 🟠 The entity remains unmanageable. No correct/refine, retain-without-use, archive, merge or feedback action.
3. 🟡 No-examples does not explain a possible count mismatch. A generic Empty is insufficient provenance.
4. 🟡 Secondary domain fetch has no failure representation. The breadcrumb can remain an ellipsis while the rule body loads.

Skill feedback

• Validates Act §3.9 correction/control and Read §1.1 compound fetches.