Content
92%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 high-quality skill body: fully executable guidance across three extraction methods, an explicitly gated verify stage, and best-practice progressive disclosure with a per-stage loading schedule. The only real cost is token budget — roughly 470 lines with a few redundant summary sections and some Stage 1 detail duplicated in the extraction reference.
Suggestions
Delete the 'When to use this skill' section (it restates the frontmatter description) and the 'When not to apply' pointer (the end-of-file 'When NOT to Use This Skill' section is already discoverable via the table of contents).
Trim Stage 1's inline JS snippets to one representative extractor and defer the rest to references/extraction.md, which already advertises the 'full JS extraction toolkit'.
Move the 'What to Extract' priority list into Stage 1 where it is acted upon, rather than placing it after the Verify stage at the end of the file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes competence — commands and code dominate, and the anti-pattern sections document genuinely non-obvious gotchas (Tailwind v4 hsl() double-wrap, hover-state extraction, prefers-color-scheme) rather than re-teaching known concepts. It falls short of the top anchor because 'When to use this skill' restates the description, 'When not to apply' is a redundant forward-pointer to the end-of-file section, and Stage 1 inlines three JS snippets that partially duplicate the extraction reference. | 4 / 5 |
Actionability | Everything is copy-paste executable: agent-browser commands with artifact paths, heredoc JS extraction payloads (with fallbacks for empty CSS-var and SPA cases), curl/grep pipelines for Method B, a complete Tailwind v4 index.css block, and before/after TSX for token migration. Specific examples cover the common cases across all three extraction methods. | 5 / 5 |
Workflow Clarity | Four explicitly sequenced stages (Extract → Document → Apply → Verify) with a method-selection decision tree and a hard validation gate — "the workflow is not complete until a diff screenshot confirms the local build matches the reference site" — plus symptom→fix error-recovery patterns delegated to the verification reference. The sequence and checkpoints are explicit, matching the top anchor. | 5 / 5 |
Progressive Disclosure | Structure is exemplary: a Reference Loading Schedule table that says what to load and what NOT to load at each stage, three references that all exist on disk, are one level deep, and are clearly signaled with priority labels (CRITICAL/HIGH) and stage timings. The SKILL.md serves as an overview while bulk detail lives in the references. | 5 / 5 |
Total | 19 / 20 Passed |