Content
82%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-engineered instruction skill: the body stays an actionable overview with hard gates, exact commands, and file conventions, while details live one link deep in a well-organized 34-file bundle. Remaining gaps are small: a missing output-verification step in the workflow, slight repetition of policy caveats, and one orphaned bundle file.
Suggestions
Add an explicit post-generation validation step in Step 4 (e.g., verify cover.png exists and its dimensions match the confirmed aspect ratio) and pair the auto-retry with a brief diagnose-and-adjust loop.
Either link references/base-prompt.md from the body (e.g., from Step 3 or the prompt-template reference) or remove it, so no bundle content is unreachable from SKILL.md.
Drop the duplicated tool-substitution caveats (stated in both 'User Input Tools' and 'Image Generation Tools') into one line to save tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and skill-specific throughout — tables for options, dimensions, and file structure; no explanations of concepts Claude already knows — e.g., the backend resolution list and "⛔ Raster only" / "⛔ No bitmap text patching" rules state policy in minimal tokens. Minor trimming is possible (the tool-substitution caveat is stated twice: "Concrete AskUserQuestion references below are examples" and "Tool names above are examples; use the local equivalent in other runtimes"; the Confirmation Policy is restated at Step 2), which keeps it at anchor 4 rather than 5. | 4 / 5 |
Actionability | Guidance is fully executable for an instruction-only skill: exact commands ("Skill with skill: 'imagegen'", "baoyu-image-gen --provider codex-cli", "test -f refs/ref-NN-{slug}.{ext}"), exact file-naming patterns ("prompts/NN-{type}-[slug].md"), concrete directory layouts, and a copy-ready completion-report template. Anchor 5's 'specific examples cover the common cases' is met, with prompt details correctly delegated to workflow/prompt-template.md. | 5 / 5 |
Workflow Clarity | Steps 0-5 are clearly sequenced with a progress checklist, explicit gates ("⛔ BLOCKING unless --quick", "Hard gate per the Confirmation Policy"), a backup-before-regenerate rule, and an error-recovery path ("On failure: auto-retry once") — matching anchor 4's 'clear sequence with most checkpoints present'. It falls short of anchor 5 because there is no post-generation verification step (e.g., confirming cover.png exists / matches the requested aspect) and the retry loop has no diagnose-and-fix phase. | 4 / 5 |
Progressive Disclosure | The body is a genuine overview: every detail domain is one clearly-signaled link away (types.md, palettes/, renderings/, dimensions/*.md, workflow/*.md, config/*.md), all 18 referenced paths resolve to real files, and cross-links between reference files are navigational rather than content-burying. One organization gap keeps it from anchor 5: references/base-prompt.md (~100 lines of base-prompt scaffold) is orphaned — linked from neither SKILL.md nor any other bundle file. | 4 / 5 |
Total | 17 / 20 Passed |