Content
75%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.
A well-structured, dense instruction skill: concrete API and schema guidance, exact design tokens, a clear six-step workflow, and details properly deferred to a references/ bundle. The main weaknesses are mild redundancy between the layout and visual-rules sections (and with the frontmatter description), a resource map that does not exactly match the real bundle files, and validation appearing only as a terminal self-check rather than inline checkpoints.
Suggestions
Correct the resource map to match the actual bundle: remove or actually add the missing root-level example.html, and list references/README.md.
Add an inline validation checkpoint between data collection and artifact writing (e.g., verify all required metrics/contributors fields are populated and label estimates before rendering).
Trim redundancy: collapse the duplicated sidebar/KPI-card guidance in 'Lay out the page' and 'Visual rules' into one section, and shorten 'When to use this skill' since the frontmatter description already carries the triggers.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes Claude's competence — exact API endpoints and response fields, exact hex/px design tokens, no concept explanations. Minor trimming is possible: the 'When to use this skill' section largely repeats the frontmatter description, and the 256px sidebar / KPI-card counts appear in both 'Lay out the page' and 'Visual rules'. This matches anchor 4 ('efficient; minor instances of over-explanation that could be trimmed') rather than the fully lean anchor 5. | 4 / 5 |
Actionability | Guidance is highly concrete: named endpoints ('GET /repos/{owner}/{repo}/contributors?per_page=100&page=N'), exact response fields, a full normalized data schema (required keys, per-row fields), exact CSS values ('0 1px 2px rgba(10,10,10,.04)...'), and named output files. It falls short of anchor 5 only in lacking a copy-paste-ready example command or HTML snippet — execution leans on references/template.html, which the body points to rather than inlines ('use the Soft Paper defaults reflected in references/template.html'). | 4 / 5 |
Workflow Clarity | The six-step workflow is clearly sequenced with concrete edge-case handling (ask for owner/repo if missing, label first-page estimates, mark synthetic deltas in provenance) and a 'Self-check' checklist at the end. Anchor 5 requires explicit inline validation steps with error-recovery feedback loops; here validation only appears as a terminal checklist with no mid-flow checkpoint (e.g., verifying fetched data before writing the artifact), so anchor 4 ('clear sequence with most checkpoints present; minor validation gaps') is the best fit. The destructive/batch cap at 3 does not apply: the operations are read-only data collection plus writing new artifact files. | 4 / 5 |
Progressive Disclosure | The body is an appropriate ~90-line overview with a resource map, clearly sectioned (workflow / visual rules / self-check), and the detailed materials are correctly pushed one level deep into references/ (template, example data, artifact and provenance examples) — all of which exist in the bundle. Two minor gaps keep it from a 5: the resource map lists 'example.html ← rendered reference dashboard' at the skill root, which does not exist in the actual bundle, and the existing references/README.md is not mentioned; references are shown in a tree diagram rather than clearly signaled markdown links. | 4 / 5 |
Total | 16 / 20 Passed |