Conciseness

Lean and dense with no over-explanation of concepts Claude already knows; length is justified by a genuinely complex multi-phase process, and every rhetorical line (e.g. the lab-notebook framing) encodes a load-bearing invariant rather than padding, matching the score-5 anchor over the 4 ('minor instances that could be trimmed').

Actionability

Fully executable guidance throughout — concrete ToolSearch calls ('ToolSearch select:LSP', the full mcp__claude-in-chrome select list), an LSP-recipe table (workspaceSymbol/documentSymbol/goToDefinition/findReferences), named output files, a copy-paste report template block, and specific grep checks — matching the score-5 'copy-paste ready' anchor.

Workflow Clarity

Six sequenced phases (ANCHOR→CLOSE) each with an explicit gate (G0–G5), feedback loops ('If neither A nor B is reachable, ask the operator', 'If ToolSearch returns none...report G4 as UNVERIFIED'), an annotation audit and recognition check, plus a scan-before-ship Hard Rules checklist — matching the score-5 anchor with explicit validation and error-recovery loops; the destructive/batch cap does not apply because validation is pervasive.

Progressive Disclosure

SKILL.md is an overview that keeps essential tables/gates/procedure inline and references seven one-level-deep files for full recipes (harvest-playbook, control-derivation, anatomy, fidelity-tiers, verification, ui-fidelity, exemplar-visit-report), all verified to exist, clearly signaled inline and enumerated in a References section — matching the score-5 'clear overview with well-signaled one-level-deep references' anchor.

Specificity

Lists multiple concrete actions spanning the full lifecycle — 'produces a single self-contained HTML prototype', 'Rebuilds existing screens at high fidelity from real source', 'derives one control per acceptance criterion', 'drives real data through a dev-server proxy', 'verifies every state in a real browser', 'reports what the spike settled' — which is comprehensive coverage, matching the score-5 anchor rather than the 4 (which allows minor gaps).

Completeness

Explicitly answers both what ('produces...HTML prototype that doubles as a design spike' plus the rebuild/derive/drive/verify/report actions) and when ('Use when a requirement or an unproven technique must be made clickable before it is built' plus a 'Triggers on...' clause), matching the score-5 anchor with concrete trigger phrases.

Trigger Term Quality

Good keyword coverage with four explicit natural triggers — 'prototype FR-NNN', 'clickable prototype of FR-NNN', 'make this requirement clickable', 'spike this technique' — but a few plausible natural synonyms (e.g. 'interactive prototype', 'mockup', 'clickable mockup') are absent, so it sits at 4 rather than the comprehensive-synonym bar of 5.

Distinctiveness Conflict Risk

A clear niche (clickable requirement/technique spikes) reinforced by explicit boundary NOTs — 'NOT for inventing UI from a prompt', 'NOT for production frontend code', 'NOT the client-confirmation record', 'NOT an ADR writer' — yielding minimal conflict risk, matching the score-5 anchor.