Content
86%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 lean, well-structured instruction skill: concrete constraints, a clear implicit workflow, and a genuine itemized acceptance checklist that acts as a validation gate. The only soft spots are a few abstract principle statements that could be made operational, and the workflow being implied by section order rather than explicitly numbered.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~21-line body is dense with directives and free of padding: it never explains concepts Claude already knows, never restates the obvious, and every clause adds a constraint or a decision rule (e.g., "先锁定用户给定的年龄、体态、脸部比例、发际线和已存在的辨识点，再写发型、服装剪裁、材质、配饰及穿用状态"). This matches 'lean and efficient; assumes Claude's competence; every token earns its place' — it is even terser than the anchor examples. | 5 / 5 |
Actionability | The guidance is concrete and executable: it specifies which attributes to lock first (age, body type, facial proportions, hairline, existing identifying marks), which material properties to describe (fabric thickness, weave, seams, stress wrinkles, justified wear positions), hard constraints per image type (soft neutral lighting, no subtitles/labels, consistent left/right accessories), and an itemized acceptance checklist ("年龄和选角是否贴合角色；脸与眼睛是否自然；头发是否像假发..."). Per the scoring notes, absence of code is not penalized in an instruction-only skill. It falls short of 5 because several statements are abstract principles rather than executable checks (e.g., "所有细节服从项目视觉风格", "材质与维护服从年代、身份和具体生活方式") that leave the reader to translate them into specific actions. | 4 / 5 |
Workflow Clarity | The section order encodes a clear sequence — lock identity attributes first, then styling/materials, then the two image-type purposes, then acceptance — and the final section is an explicit per-item verification gate that even guards against a false-positive ('文字规范通过或图片任务完成，都不等于视觉合格'). It matches 'clear sequence with most checkpoints present; minor validation gaps': the steps are implied by section order rather than numbered, and there is no explicit fix-and-recheck loop after the acceptance check, which keeps it below 5. | 4 / 5 |
Progressive Disclosure | This is a single-file skill under 50 lines with no references/, scripts/, or assets/ directories, and the rubric's simple-skill exception applies: progressive disclosure can score 5 with just well-organized sections. The body has clean, well-ordered headers (身份与妆造, 两类图的用途, 看实际图再验收) that keep the whole skill scannable in one pass with no nested references to chase. | 5 / 5 |
Total | 18 / 20 Passed |