Content
77%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
Highly actionable and easy to navigate per-section, with complete executable code throughout. Its weaknesses are token efficiency (restating well-known Three.js property APIs inline) and the absence of any progressive disclosure — everything lives in one long SKILL.md with no reference files.
Suggestions
Move the exhaustive MeshPhysicalMaterial property listing and the Common Material Properties block into a references/ file (e.g., references/physical-material.md), keeping SKILL.md to the overview table, quick start, and one concise example per material type.
Remove sections that restate knowledge Claude already has, such as "Built-in Uniforms (auto-provided)" for ShaderMaterial, and keep only the non-obvious notes (e.g., the aoMap uv2 requirement, computeLineDistances for dashed lines).
Convert the inlined bulk listings into well-signaled one-level-deep references (e.g., "**Physical material properties**: See [PHYSICAL.md](references/physical.md)"), matching the See Also pointers to sibling skills.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is code-dense with almost no padded prose, but sections like "Common Material Properties" and "Built-in Uniforms (auto-provided)", plus the exhaustive MeshPhysicalMaterial property dump, restate well-known Three.js API knowledge Claude already has; the ~520-line file could be tightened considerably. | 3 / 5 |
Actionability | Every material type has complete, copy-paste-ready instantiation code with inline comments, plus concrete recipes (glass, car paint, gradient map, dashed lines, material pooling) that cover the common cases. | 5 / 5 |
Workflow Clarity | This is a single-action reference skill rather than a multi-step workflow: the material-types table unambiguously guides selection and each section shows the full instantiation step, and there are no destructive or batch operations that would require validation checkpoints. | 5 / 5 |
Progressive Disclosure | Internal structure is good (quick start, clear headers, overview table), but there are no bundle files at all — the ~250 lines of exhaustive property listings belong in reference files (e.g., a physical-material properties reference), leaving SKILL.md as a monolithic lookup page. | 3 / 5 |
Total | 16 / 20 Passed |