Conciseness

The ~310-line body is noticeably verbose: the 'journey is not a third delivery lane' rule is stated three times (section 2, Common Mistakes, Pressure Test), and the workspace-claim paragraph embeds repo-specific gate policy with time-sensitive dates ("2026-09-10 起，静态 prover 已被 runtime journey 取代", "S2 #4669 起") that are not in a deprecated section. It is not anchor 1 because nearly all content is genuine domain judgment knowledge rather than concepts Claude already knows.

Actionability

Guidance is largely concrete and executable: named commands ("pnpm check:design-gate-real-interaction"), file paths ("packages/web/test/browser/", "docs/design-gate-claims/<id>.json"), deterministic verdict enums (SUPPORTED/REFUTED/UNKNOWN), and a stepwise sentinel-test checklist. It falls short of anchor 5 because some sections remain at judgment-rule level without executable specifics, but well above anchor 3's pseudocode-level guidance.

Workflow Clarity

A clear numbered sequence (先路由 → §1 lock the judging question → §2 delivery lane → … → §6 Demo Contract → §7 build → §8 verification) with explicit validation checkpoints: the per-claim verification-mechanism table, the first-frame check, and the Pressure Test table with failure signals that gates entry into frontend implementation ("任一场景若无法从 Demo Contract 直接读出判题类型、交付车道、视觉真相源和诚实边界，不进入前端实现"). This matches anchor 5's clear sequence plus explicit validation steps, error-recovery feedback (Common Mistakes 修正 column), and checklists.

Progressive Disclosure

References are clearly signaled one level deep ("refs/demo-contract-template.md", "refs/falsifiable-technical-cutaway.md", "refs/lessons-from-two-demos.md", "refs/interaction-narrative-cases.md") but none of these files exist in the provided bundle, so the links cannot be verified or loaded. Additionally, deep case policy (the workspace-claim checker/runtime-gate mechanics paragraph) is inlined in SKILL.md where it clearly belongs in a reference file. This fits anchor 3 — some structure, references present and signaled, but content that should be separate is inline — rather than anchor 2's minimal/buried structure or anchor 4's well-placed content.

Specificity

The description names concrete actions ("拆解参考产品的交互叙事，或把理念与用户旅程变成可判断的 Demo") and enumerates deliverables ("双轴 Demo Contract、交互原型与验证记录"). It lists several specific actions with minor gaps in coverage, matching anchor 4 rather than the comprehensive coverage of anchor 5, which would require a fuller action inventory.

Completeness

It explicitly answers both what ("拆解参考产品的交互叙事，或把理念与用户旅程变成可判断的 Demo") and when ("Use when: ...") with concrete trigger phrases, and goes further with explicit "Not for" exclusions and an "Output" statement. This clearly matches anchor 5; anchor 4 would require the 'when' to be less explicit, which it is not.

Trigger Term Quality

The "Use when" clause contains natural phrases a user would actually say ("参考产品设计怎么借鉴、产品引导的叙事节奏、做个 demo 让人 get 到、比较家内交互或验证完整旅程"). Coverage is good but a few natural variations and synonyms are missing, which fits anchor 4 rather than the comprehensive synonym coverage of anchor 5 or the partial coverage of anchor 3.

Distinctiveness Conflict Risk

The "Not for" clause ("已签字的正式产品前端、已有素材剪辑、PPT、纯视觉探索、用户实际操作引导") carves out a clear niche with distinct triggers, explicitly disambiguating from adjacent production/video/PPT/guide skills. Minimal conflict risk — anchor 5.