Content
36%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 skill body reads as a capability catalog: it documents features, algorithms, and safety boundaries with genuinely useful domain data (paths, MET values, danger thresholds), but it explains statistics Claude already knows, inlines three full report transcripts, and never sequences an end-to-end analysis workflow. It would benefit most from a lead workflow, trimmed boilerplate, and moving reference material into separate files.
Suggestions
Lead with a numbered workflow (read data from the named paths → run quality checks with explicit pass/fail handling → compute trends/correlations → render report from a single compact template → apply danger-signal checks before finalizing) so the process is sequenced with validation checkpoints.
Cut the explanations of basic statistics (regression slope sign, Pearson r interpretation) and keep only the domain-specific thresholds, strength cutoffs, and MET table; consolidate the three full example reports into one annotated template.
Move the report templates, algorithm details, and medical safety reference into files under references/ and link them from a short SKILL.md overview, enabling progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~430-line body pads token budget with material Claude already knows: linear regression basics ('b > 0：上升趋势'), Pearson correlation interpretation ('r = 1：完全正相关... r = 0：无线性相关'), and three full fabricated report transcripts (trend, correlation, progress) whose value could be conveyed by one compact template. This matches 'noticeably verbose; several unnecessary explanations or padded sections' — not 3 because the amount of redundant explanation and full-report padding is substantial, though the MET table, thresholds, and data paths do earn their place. | 2 / 5 |
Actionability | There is real concrete guidance — data source paths like 'data/fitness-logs/YYYY-MM/YYYY-MM-DD.json', MET calorie formula with activity values, pace calculation, and slash-command examples ('/fitness trend 3months') — but no executable code or scripts for the regressions/correlations, and the slash commands are not tied to how the skill is actually invoked. This matches 'some concrete guidance but incomplete; pseudocode instead of executable code; missing key details'. | 3 / 5 |
Workflow Clarity | The body is organized by capability (features, output formats, data sources, algorithms, safety) rather than as a sequence from user request to finished report. Data-quality checks are listed as bullets, not as validation checkpoints in a workflow, and there are no error-recovery steps. This matches 'rough sequence present but many gaps; steps poorly defined; validation absent' — better than 1 because the section ordering roughly implies a flow. | 2 / 5 |
Progressive Disclosure | The single file has clear, well-labeled sections (## 功能, ## 输出格式, ## 数据源, ## 算法说明, ## 医学安全边界) making it navigable, but everything is inlined in one ~430-line SKILL.md with no bundle files or references — the report templates, algorithm notes, and MET table are content that could live in separate reference files. This matches 'some structure but could be better organized... content that should be separate is inline'. | 3 / 5 |
Total | 10 / 20 Passed |