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.
A highly actionable, well-sequenced skill body that teaches a proprietary API with executable examples, explicit error handling, and verification checkpoints. Its main weakness is redundancy — repeated distinctions, duplicated quick-reference tables, and a worked example that re-narrates the steps — which inflates token cost without adding information.
Suggestions
Consolidate the INSTANCE_SWAP-vs-SLOT guidance (currently stated in Step 3, the property-mapping table, the SLOT section, the descendants table, and Rule 5) into one authoritative location and cross-reference it.
Trim the Inline Quick Reference to method names plus one-line notes and defer full signatures and SelectorOptions examples to references/api.md, removing duplication with the property-mapping tables.
Compress the Complete Worked Example to the final Button.figma.ts template with brief step annotations instead of re-narrating Steps 1-6, and fix the duplicate numbering of two pitfalls as '7' in Rules and Pitfalls.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly dense, proprietary-API knowledge Claude genuinely lacks (ErrorHandle truthiness, ResultSection interpolation, exhaustiveness requirements), but there is more than minor fat: the INSTANCE_SWAP-vs-SLOT distinction is restated in at least five places (Step 3, the property-mapping table, the SLOT section, the descendants table, and Rule 5), the Inline Quick Reference (~70 lines) duplicates the earlier property-mapping tables and references/api.md, the Complete Worked Example re-narrates all six steps verbatim, and two pitfalls share the number 7. This fits 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than anchor 4's 'minor instances'. | 3 / 5 |
Actionability | Fully executable guidance throughout: exact MCP tool invocations with named parameters ("get_code_connect_suggestions with fileKey... nodeId... excludeMappingPrompt: true"), copy-paste-ready template code with WRONG/CORRECT contrasts, concrete URL-parsing rules with a worked example, and a complete end-to-end Button.figma.ts example covering the common case. Matches the anchor-5 'copy-paste ready code; specific examples cover the common cases'. | 5 / 5 |
Workflow Clarity | A clear six-step sequence with explicit validation and feedback loops: Step 2 handles each error response verbatim ("No published components found in this selection" → inform and stop), Step 4 requires user confirmation before proceeding, and Step 6 is a read-back validation checklist (property coverage, typed code, no hardcoded children, interpolation wrapping). This matches anchor 5's 'explicit validation steps; feedback loops for error recovery; checklists'. | 5 / 5 |
Progressive Disclosure | Good structure with real, verified, one-level-deep references — [api.md](references/api.md) carries the full API reference and [advanced-patterns.md](references/advanced-patterns.md) the advanced nesting patterns, both clearly signaled at point of need. It stops short of anchor 5 because reference-appropriate content is inlined: the full method-signature tables in the Inline Quick Reference overlap api.md, and the 520-line body keeps the complete worked example inline rather than splitting it out — fitting 'most content is appropriately placed; minor organization gaps'. | 4 / 5 |
Total | 17 / 20 Passed |