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 exceptionally actionable, well-sequenced workflow: mandatory ordered steps, executable scripts with exact API shapes, explicit validation checkpoints (structural read-back evidence, one composition screenshot, font assertion), and error-recovery guidance, with genuine one-level-deep references. Its weakness is token efficiency — the hard gate, parallel-workflow caveats, and a full Best Practices recap each restate rules already specified in the steps — and a few extended patterns are inlined that could live in reference files.
Suggestions
Delete or drastically prune the Best Practices recap: nearly every bullet restates rules already specified in Steps 2–5 (search-first, one-intent queries, token preference, componentize by default, retry-safe phases, node-ID returns, screenshot cadence, font assertion), reclaiming a whole section of duplicated tokens.
State the editable-layers/screenshot-prohibition gate once at the top of the body and cross-reference it from the Parallel Workflow section and Step 5 instead of repeating it four times.
Move extended pattern details — the icon SVG import procedure, the generate_figma_design image-transfer steps, and library pagination mechanics — into reference files alongside componentization.md to shorten the main body and strengthen progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is information-dense — nearly every paragraph carries non-obvious operational facts (appendChild silently failing across calls, currentColor importing as black, library pagination, variable-API blind spots) — but it is noticeably redundant: the editable-layers/screenshot prohibition is stated at the top, again in the Parallel Workflow section, and twice more in Step 5; the fileKey/create_new_file guidance appears in both Prerequisites and the Parallel Workflow; and the ten-bullet Best Practices recap restates rules already specified in Steps 2–5. This fits 'mostly efficient but includes some unnecessary explanation or could be tightened' (3) better than 'minor instances of over-explanation' (4) — the recap is a whole section of duplication. | 3 / 5 |
Actionability | Fully executable throughout: complete use_figma scripts for component-key resolution, the existing-screen inventory, wrapper sizing by container type, variable-bound section construction, SVG icon import (with the width/height viewBox gotcha), imageHash transfer, and variant swapping; plus exact search_design_system query shapes, a worked fileKey/nodeId parsing example, and a build-manually-vs-import decision table. Copy-paste-adaptable code covers the common cases — top anchor. | 5 / 5 |
Workflow Clarity | "Follow these steps in order. Do not skip steps." six ordered steps with a required-invariants block, explicit validation checkpoints (read-back descendant counts as structural evidence, exactly one composition screenshot plus one post-fix shot, mandatory font-family assertion, and the hard gate: rebuild semantically or report failure), and a dedicated Error Recovery section with retry semantics. Clear sequence, explicit validation, and feedback loops — top anchor. | 5 / 5 |
Progressive Disclosure | Both local bundle files (references/componentization.md, references/discover-product-font.md) exist and are well-signaled with purpose statements, and the Reference Docs section cleanly delegates API patterns to the figma-use bundle one level deep. However, the ~470-line body inlines several extended patterns (icon SVG import details, the image-transfer procedure, library pagination mechanics) that could be split into reference files, matching 'good structure; most content appropriately placed; minor organization gaps' (4) rather than 'content appropriately split' (5). | 4 / 5 |
Total | 17 / 20 Passed |