Content
50%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.
The skill's core usage documentation is genuinely actionable, with executable commands for all four study types and sensible smart-inference defaults. However, roughly half the body is generic process boilerplate, and structural defects (a dangling 'See ## Usage above' reference, duplicate contradictory validation shortcuts, a leaked internal path `cd "20260316/scientific-skills/Others/clinic-sample-size"`, and a title appearing mid-document) undermine both conciseness and workflow clarity.
Suggestions
Delete the generic template sections ('Output Contract', 'Deterministic Output Rules', 'Completion Checklist', 'Validation and Safety Rules', 'Failure Handling', 'Required Inputs') and keep only skill-specific content; the ~90 lines of boilerplate are the main token waste.
Consolidate the two validation shortcuts into one that actually works (validate_skill.py), remove the incorrect `python scripts/calculators.py --help` claim and its fabricated 'Expected output format', and fix the 'See `## Usage` above' dangling reference in Example Usage.
Restructure so the title '# Clinic Sample Size Calculator' leads the document, followed by When to Use / Not to Use and the Usage section, and remove the leaked internal build path from the example commands.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | A large fraction of the ~188-line body is generic template filler that adds no skill-specific information: "Use this skill when the request matches its documented task boundary", "Output discipline: keep results reproducible, identify assumptions explicitly, and avoid undocumented side effects", "Return a structured deliverable that is directly usable without reformatting". Sections like 'Output Contract', 'Deterministic Output Rules', 'Completion Checklist', and 'Validation and Safety Rules' restate process boilerplate Claude already knows. Not 1 because the core Usage section with the four study-type commands is genuinely useful and unpadded; not 3 because the filler is extensive, roughly half the document. | 2 / 5 |
Actionability | The Usage section provides concrete, copy-paste-ready commands covering all four study types, e.g. `python scripts/main.py --study_name "New Biomarker Study" --outcome "Sensitivity > 0.8" diagnostic sens_spec --se 0.8 --sp 0.9 --error 0.05 --prev 0.3 --dropout 0.1`, plus inference-mode examples with commented default values. Not 5 because of gaps: the 'Quick Validation' documents `python scripts/calculators.py --help`, but calculators.py is a module with no CLI, and the efficacy section only shows one of the documented input variants. Not 3 because the primary paths are fully executable and cover the common cases. | 4 / 5 |
Workflow Clarity | Sequences exist ('Example run plan', 'Recommended Workflow') and validation is attempted ('Validation Shortcut', 'Quick Validation'), but the checkpoints are muddled: two different 'run this first' commands are given (validate_skill.py vs calculators.py), the 'Expected output format' under Quick Validation describes report output a --help call would never produce, and 'Example Usage' opens with "See `## Usage` above" pointing at a section that appears later in the document. Sequence present but checkpoints are duplicated/confusing rather than clear — matches anchor 3, not 4. | 3 / 5 |
Progressive Disclosure | The bundle is scripts-only (main.py, calculators.py, validate_skill.py — all real and correctly referenced), and per-study-type details appropriately live in the scripts rather than SKILL.md. However, structure within SKILL.md is disorganized: the actual title '# Clinic Sample Size Calculator' appears midway into the document, 'When to Use' precedes it, and there are two competing validation sections. This fits 'some structure but could be better organized' — not 4, since the interleaved boilerplate and mid-file title make navigation harder than the 'good structure' anchor. | 3 / 5 |
Total | 12 / 20 Passed |