Conciseness

The body is dense and high-signal, pinned to r186 with deprecations (r180–r183) correctly isolated in the Common Mistakes section acting as an old-patterns list. It falls short of anchor 5 because the Quick Start and WebGPU sections repeat renderer/camera boilerplate Claude already knows, and "Three-point lighting: key, fill, rim" restates a familiar concept. It is above anchor 3 because nearly every line carries version-specific or unit-specific facts rather than padding.

Actionability

Every section provides fully executable, copy-paste-ready code covering the common cases: light setup, shadow tuning with concrete values ("dirLight.shadow.normalBias = 0.02; dirLight.shadow.bias = -0.0001"), IBL assignment, probe baking, and WebGPU registration lines. This matches the fully-executable anchor and is not anchor 4 since there are no pseudocode or missing-detail gaps.

Workflow Clarity

Multi-step procedures are clearly sequenced — the shadow setup is numbered (renderer → light → objects) with three explicit switches, and shadow-camera tuning gives ordered, concrete parameters. It is not anchor 5 because tuning workflows (e.g., shadow acne) lack a diagnostic feedback loop ("if acne persists, increase normalBias"), though no destructive or batch operations exist to cap the score at 3.

Progressive Disclosure

The body is well-organized with clear headers, but there is no progressive disclosure at all: no references/, scripts/, or assets/ files exist, and separable material (WebGPU/TSL specifics, light probe grids, per-light reference details) is inline in a single ~480-line file. This matches anchor 3 (good inline structure, but content that should be separate is inline) rather than anchor 4, which expects most bulk content placed in separate referenced files.

Specificity

The description lists concrete capabilities comprehensively — "Light types and physical units, shadow maps, HDR environment lighting, light probes, and lighting performance, on WebGL or with WebGPU and TSL" — with no vague filler. It matches the comprehensive-coverage anchor rather than the minor-gaps anchor because the topic list fully spans the skill's scope.

Completeness

It explicitly answers both what ("Light types and physical units, shadow maps, HDR environment lighting, light probes, and lighting performance") and when ("Use when adding or tuning lights, fixing shadow acne, setting up image-based lighting, or dimming an environment map") with concrete trigger phrases. It is not anchor 4 because the 'when' clause is fully explicit rather than only partially specific.

Trigger Term Quality

Trigger phrases like "adding or tuning lights", "fixing shadow acne", "setting up image-based lighting", and "dimming an environment map" are natural user phrasings. It falls short of anchor 5 because common synonyms users would say — e.g. "point light", "spotlight", "directional light", "shadows too dark" — are missing.

Distinctiveness Conflict Risk

Lighting (lights, shadow maps, IBL, probes) is a distinct niche from sibling material/texture/postprocessing skills, with distinct triggers like "shadow acne". It is not anchor 5 because the description never names three.js — only the jargon "TSL" and "WebGL" hint at it — so generic terms like "lighting performance" could pull in lighting questions for other frameworks.