Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The content is highly actionable with clear phased workflows, but it is verbose and monolithic: it explains GoF material Claude already knows and inlines reference content while pointing to bundle files that are not present. Tightening and actually externalizing the referenced material would raise the score.
Suggestions
Move the detection rules, code-smell mappings, stack-pattern equivalents, and evaluation checklists into the referenced `signatures/` and `checklists/` files and ship them, so SKILL.md becomes a lean overview rather than inlining that material.
Trim the large illustrative JSON/Markdown report examples (Detection/Suggestion/Evaluation modes) to minimal skeleton schemas, removing padding that does not change the instruction.
Cut the inline 23-pattern catalog and code-smell definitions, which Claude already knows, and keep only the stack-specific detection heuristics and adaptation logic that are non-obvious.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~550-line body restates concepts Claude already knows (the full 23-pattern catalog, code-smell definitions, SOLID criteria) and includes very large illustrative JSON/Markdown report examples; it is mostly efficient but could be tightened considerably and is not lean enough for a 3. | 2 / 3 |
Actionability | It provides concrete, executable guidance — specific grep regexes (e.g. `switch (.*type)`, `function \w+\([^)]{60,}\)`), real TypeScript strategy/Zustand examples, and exact CLI invocations — making it copy-paste ready rather than abstract. | 3 / 3 |
Workflow Clarity | Each operating mode and Phase 1–5 methodology is laid out as an explicit numbered sequence with confidence/scoring gates; the skill is read-only analysis, so the destructive/batch validation requirement does not apply. | 3 / 3 |
Progressive Disclosure | References to `signatures/`, `reference/`, and `checklists/` files are signaled and one level deep, but those bundle directories do not exist, and the detection rules, code smells, stack mappings, and evaluation criteria are inlined in a monolithic SKILL.md rather than split out — matching the anchor where content that should be separate is inline. | 2 / 3 |
Total | 10 / 12 Passed |