Content
81%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 body is an exemplar of executable, well-gated workflow design: ordered steps, hard gates, per-section validation, a font assertion, and an atomic error-recovery loop, all backed by copy-paste-ready scripts. Its weaknesses are redundancy — the componentization section and Best Practices recap content stated elsewhere or duplicated with the reference file — and cross-bundle reference links that do not resolve within the skill directory.
Suggestions
Deduplicate the componentization section: the SKILL.md paragraphs repeat references/componentization.md nearly verbatim — keep a one-line summary plus the link, and let the reference carry the rationale.
Trim the Best Practices section, which restates rules already enforced in Steps 2–5 (search-before-build, one intent per query, tokens over hardcoded values, componentize by default); keep only the points not already covered.
Consolidate the repeated fileKey / create_new_file instructions (given in both Prerequisites and the Parallel Workflow section) into one place, and consider moving the detailed icon SVG-import subsection into a reference file to slim the main body.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The bulk is high-value, non-inferable operational detail, but there is real duplication: the componentization paragraphs ("Componentization is part of the **default** workflow, not an optional follow-up... do not emit a flat tree of one-off frames") are restated nearly verbatim from references/componentization.md, the Parallel Workflow section repeats the fileKey/create_new_file guidance already given in Prerequisites, and Best Practices recaps five rules already stated in Steps 2–5. This matches the 3 anchor (mostly efficient, some unnecessary content that could be tightened) better than the 4 anchor, whose over-explanation is only minor. | 3 / 5 |
Actionability | Nearly every step ships copy-paste-ready executable code with clearly marked placeholders: the Code Connect key resolver ("figma.getNodeByIdAsync('609:35535')... return { componentKey: set.key }"), the wrapper sizing script, the Promise.all section-build script, the SVG icon import ("figma.createNodeFromSvg('<svg width="24" height="24" viewBox=...>')"), and the imageHash transfer loop. Query strategies give concrete term lists ("gray", "background", "space", "radius"). This is fully executable and covers the common cases per the 5 anchor. | 5 / 5 |
Workflow Clarity | The workflow is an explicitly ordered 6-step sequence ("Follow these steps in order. Do not skip steps.") with hard gates forbidding shortcuts, per-section validation via get_screenshot ("Look closely for cropped/clipped text... and overlapping elements"), a mandatory font assertion in Step 5, and a full error-recovery feedback loop (STOP → read error → inspect state → fix script → retry, with an atomicity guarantee). This matches the 5 anchor's explicit validation steps, feedback loops, and checklists. | 5 / 5 |
Progressive Disclosure | Structure is good: a well-signaled "Reference Docs" section with one-line descriptions per file, and local references (references/componentization.md, references/discover-product-font.md) that exist and are one level deep. However, most reference links point outside the bundle (../figma-use/references/component-patterns.md, variable-patterns.md, text-style-patterns.md, effect-style-patterns.md, gotchas.md, ../figma-generate-library/references/component-creation.md) — paths that do not resolve within this skill's directory — and detailed material like the full icon/SVG subsection is inlined rather than split out. That places it at the 4 anchor (good structure, minor organization gaps) rather than 5. | 4 / 5 |
Total | 17 / 20 Passed |