Content
21%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill reads more like a TCM textbook chapter than an actionable skill for Claude. The vast majority of content is encyclopedic reference material (constitution descriptions, food lists, acupoint locations, herbal formulas) that should either be in separate data files or omitted entirely since Claude already has this knowledge. The actual workflow guidance is minimal and lacks validation steps, while the document's extreme length (~500+ lines) wastes precious context window space.
Suggestions
Reduce the document to under 100 lines by moving all reference content (constitution characteristics, dietary recommendations, seasonal advice, herbal formulas, example reports) into separate bundle files like `data/constitutions.json` and `templates/report.md`.
Remove explanations of TCM concepts Claude already knows (what each constitution type is, basic food properties, acupoint locations) and focus only on the specific scoring algorithm, data file formats, and output format requirements.
Add concrete validation steps to the workflow: verify questionnaire completeness (all 60 questions answered), validate score ranges (0-100), check for data file existence before reading, and verify output report completeness.
Replace the abstract trend analysis description ('linear regression', 'moving average') with executable code or specific instructions, similar to the constitution scoring algorithm that is already well-defined.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at ~500+ lines. Extensively explains TCM concepts Claude already knows (what each constitution type is, basic dietary principles, acupoint locations). The massive example report output takes up most of the document and could be a separate template file. Much content is encyclopedic reference material rather than actionable instructions. | 1 / 5 |
Actionability | The scoring algorithm code is concrete and executable, and the constitution determination logic is clear. However, much of the skill is descriptive lists (food recommendations, exercise types, acupoint descriptions) rather than executable guidance. The data file paths are specified but there's no concrete code for reading/parsing them, and the trend analysis is described only abstractly ('linear regression', 'moving average'). | 3 / 5 |
Workflow Clarity | Steps 1-4 are listed but extremely high-level ('确定分析范围', '读取数据', '数据分析', '生成报告') with no validation checkpoints. There's no feedback loop for error cases (e.g., what if data files are missing, scores are inconsistent, or questionnaire is incomplete). The workflow lacks concrete verification steps despite dealing with health-related analysis where accuracy matters. | 2 / 5 |
Progressive Disclosure | Monolithic wall of text with no bundle files to reference. The entire encyclopedic content (constitution characteristics, dietary recommendations, seasonal advice, acupoint details, herbal formulas, example reports, data structures) is inlined in a single massive file. This content desperately needs to be split into separate reference files (e.g., constitutions knowledge base, recommendation templates, example reports). | 1 / 5 |
Total | 7 / 20 Passed |