Content
56%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 core six-step workflow is well designed — sequenced, concrete, and format-driven — but it is surrounded by large amounts of generic template boilerplate that inflates token cost without adding domain value, plus internally contradictory validation guidance and a broken cross-reference. Trimming the boilerplate and reconciling the validation sections would substantially improve the skill.
Suggestions
Remove the generic template sections (Key Features, Implementation Details, Output Contract, Deterministic Output Rules, etc.) that restate the description or state obvious process rules; this would cut the file roughly in half and let the actual Workflow lead the document.
Reconcile the contradictory validation guidance: either document the script's real interface (`python scripts/validate_skill.py --check`) in one place or delete both the "Validation Shortcut" and "Quick Validation" sections.
Fix navigation errors: "See `## Workflow` above for related details" should point below (or be deleted), and drop the pipeline-specific `cd "20260316/scientific-skills/..."` path from Example Usage.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~290-line body wraps a ~165-line core workflow in heavy template boilerplate ("Key Features", "Implementation Details", "Output Contract", "Deterministic Output Rules", "Completion Checklist") that restates the description verbatim twice and adds filler like "Execution model: validate the request, choose the packaged workflow, and produce a bounded deliverable". This matches the anchor for several unnecessary padded sections; not 3 because the padding is close to half the file. | 2 / 5 |
Actionability | The workflow gives concrete, executable guidance: exact output templates, per-section field lists, and hard rules ("Core symptoms (≤5)", "18-25 characters", "Start with prefix: 【Case Report】". Not 5 because the validation-script guidance is confusing — the body documents `--help` while the script's actual `--check` flag is never mentioned, and Example Usage includes a pipeline-specific `cd` path irrelevant to users. | 4 / 5 |
Workflow Clarity | Six clearly sequenced steps each specify Input/Action/Output, backed by a completion checklist, failure handling, and input-validation rules. Not 5 because the validation guidance contradicts itself — "Validation Shortcut" says to run `python scripts/validate_skill.py --help` while "Quick Validation" says "No local script validation step is required for this skill" — and "See `## Workflow` above for related details" points to a section that appears below. | 4 / 5 |
Progressive Disclosure | A single-file monolith with no reference files; the one referenced bundle path (`scripts/validate_skill.py`) is real, but the actual workflow is buried under ~50 lines of boilerplate and a broken internal pointer. Structure exists (clear section headers) but the skill is well over 50 lines and the treatment-option and disease-intro detail could live in reference files, matching the mid anchor. | 3 / 5 |
Total | 13 / 20 Passed |