Content
50%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 contains genuinely useful, executable reference material — a real CLI with documented parameters, input formats, and example output — wrapped in a large amount of generated template boilerplate. The main quality problems are padding, broken self-references, and generic checklist sections that bloat the context window and should be moved to reference files or removed.
Suggestions
Cut the template filler sections ('Key Features', 'Evaluation Criteria', 'Lifecycle Status', 'Output Requirements', 'Response Template', and the duplicate 'Quick Check'/'Audit-Ready Commands' blocks) or move them to references/audit-reference.md, keeping SKILL.md focused on Usage, Input Format, Output, and Error Handling.
Fix the broken cross-references — 'See `## Usage` above' and 'See `## Workflow` above' point to sections defined later — and remove the non-portable 'cd "20260318/scientific-skills/..."' example in favor of a relative path from the skill root.
Ship a real requirements.txt (the Prerequisites section runs 'pip install -r requirements.txt' but no such file exists) and make the Workflow steps concrete by tying each step to the actual commands (e.g. py_compile check, then the --input/--beta/--full invocations) instead of generic directives.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is noticeably verbose with padded template sections: 'Key Features' ('Packaged executable path(s)', 'Structured execution path'), an empty 'Evaluation Criteria' checklist, 'Lifecycle Status' filler ('Planned Improvements: Performance optimization'), 'Quick Check' duplicating 'Audit-Ready Commands', two separate 'References' sections, and broken self-references ('See `## Usage` above' and 'See `## Workflow` above' both point to sections that appear later in the file). It does not hit score 1 because it avoids long explanations of concepts Claude already knows. | 2 / 5 |
Actionability | The Usage section provides fully executable commands ('python scripts/main.py --input otu_table.tsv --metric shannon --output alpha_report.html'), a concrete parameter table, exact input file formats, and a realistic example output JSON — copy-paste ready for the common cases. Minor gaps keep it below 5: 'Prerequisites' tells the user to run 'pip install -r requirements.txt' but no requirements.txt exists in the bundle, and the Example Usage section contains a non-portable 'cd "20260318/scientific-skills/..."' path plus self-referential filler ('See `## Usage` above for related details.') instead of instructions. | 4 / 5 |
Workflow Clarity | A 5-step Workflow is listed in order with an explicit fallback step, and 'Quick Check' gives one validation command (py_compile), but the steps are generic directives ('Confirm the user objective...', 'Use the packaged script path or the documented reasoning path...') rather than concrete sequenced actions with checkpoints — matching 'steps listed but validation gaps; checkpoints missing or implicit'. | 3 / 5 |
Progressive Disclosure | The bundle structure is sound — the referenced file references/audit-reference.md exists, is one level deep, and is clearly linked, and scripts/main.py exists as claimed — but the ~300-line SKILL.md body inlines large generic blocks (Risk Assessment, Security Checklist, Evaluation Criteria, Output Requirements, Response Template, audit commands) that clearly belong in the references directory, fitting 'some structure but could be better organized; content that should be separate is inline'. | 3 / 5 |
Total | 12 / 20 Passed |