Conciseness

The body is dense and imperative with no filler — lines like 'Ground penetration: compare lowest foot or shoe vertices with floor height per frame' and the quantified 'Practical Thresholds' (1-2 cm, 5%, 30 degrees per frame) earn their tokens, and it never explains basics Claude already knows. It is not 5 because the 'How It Works' section substantially duplicates the 'Inspection Workflow' steps, and 'Core Principle' restates the description — trimmable redundancy.

Actionability

Concrete guidance is strong: a runnable command ('blender --background scene.blend --python collect_motion_state.py'), specific bone names, quantified thresholds, and worked examples with measured facts ('foot.L min_z = 0.004', 'foot.L x = 0.21 -> 0.28'). It is not 5 because the central extraction step — the Blender state exporter / collect_motion_state.py script — is referenced but never provided, so the agent must author the key tooling itself; it is well above 3 because the surrounding guidance is executable and specific rather than pseudocode.

Workflow Clarity

The 7-step 'Inspection Workflow' is clearly sequenced with explicit validation gates: 'Establish the clean scene and asset baseline before judging motion', 'Check model integrity before retargeting blame', 'Report facts before opinions', plus adaptive sampling ('sample more densely around flips, landings, turns') and a report template acting as a checklist. It is not 4 because checkpoints are not merely present — they are explicit ordering principles enforced at every phase with error-recovery guidance in the anti-patterns; the read-only nature of inspection means the destructive-batch cap does not apply.

Progressive Disclosure

The skill is a single self-contained file (no references/, scripts/, or assets/ exist) with well-labeled sections (When to Use, Workflow, Report Shape, Examples, Thresholds, Anti-Patterns, Tooling Notes), so navigation is easy. It is not 5 because at ~156 lines the worked Examples and Report Shape template are natural candidates for one-level-deep reference files that would slim the main file, and no external references are signaled at all; it is well above 3 because the structure is consistent and nothing is buried.

Specificity

Quotes like 'inspecting Blender characters, rigs, poses, animation retargeting, ground contact, facing direction, or model-vs-motion alignment' enumerate several concrete, domain-specific capabilities with a discriminator condition ('where screenshots alone are not enough'). It stays at 4 rather than 5 because the capability set hangs on a single action verb ('inspecting') rather than listing multiple distinct concrete actions, and at 3 because the coverage is far more specific and comprehensive than 'names domain and 1-2 concrete actions'.

Completeness

The 'what' is explicit (inspection of Blender characters/rigs/poses/retargeting/ground contact/facing/model-vs-motion alignment) and the 'when' is an explicit 'Use this skill when...' clause with concrete trigger phrases plus a scope condition. This matches the score-5 anchor (both what and when clearly and explicitly answered with concrete triggers); it is not 4 because the 'when' is not merely implied or generic — it names the exact situations and the screenshot-insufficiency condition.

Trigger Term Quality

Natural terms a user would say are well covered: 'Blender', 'characters', 'rigs', 'poses', 'animation retargeting', 'ground contact', 'facing direction', 'screenshots'. It falls short of 5 because common variations and extensions are missing — e.g. 'foot sliding', 'mirrored', 'avatar', '.blend', 'skating' — while clearly exceeding 3's 'some relevant keywords but missing common variations'.

Distinctiveness Conflict Risk

The description carves a clear niche (Blender animation/motion-state inspection with structured-facts-vs-screenshot framing) with distinct trigger terms unlikely to fire for unrelated skills. It is not 4 because there is no meaningful overlap risk with closely related skills — the domain and discriminator ('where screenshots alone are not enough') are tightly scoped.