Content
82%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.
An exemplarily lean, well-sequenced instruction skill that delegates detail to bundle files with condition-based pointers. Its main defects are a dangling reference to a nonexistent checklist file, an unreferenced script, and slightly abstract middle steps (ownership/coupling tracing) that keep actionability and workflow validation short of the top band.
Suggestions
Create the missing "checklists/analysis-checklist.md" referenced in step 5 (or point to an existing file), since the body's only validation checkpoint depends on it.
Add one short concrete example of a confirmed finding (smell name + evidence + improvement) inline to sharpen actionability for step 4's ranking step.
Either reference "scripts/validate-metadata.py" where relevant in the workflow or remove it from the bundle to keep the file structure intentional.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~20-line body is lean and assumes competence: "Name the code smell, source evidence, and concrete improvement" states the quality bar without explaining what refactoring or code smells are. No padding or concept explanations anywhere; every token earns its place — matches anchor 5. | 5 / 5 |
Actionability | Concrete guidance throughout: exact reference paths ("references/code-smells-catalog.md", "references/refactoring-techniques.md"), a concrete output convention ("docs/_refacs/<YYYYMMDD>-<slug>.md"), ranking criteria (impact, frequency, risk, effort), and conditional handling in step 6. Falls short of anchor 5 only because steps like "Trace ownership, coupling, call sites, and public boundaries" give direction without method specifics, and no example finding is shown inline. | 4 / 5 |
Workflow Clarity | A clear six-step sequence with explicit branch handling ("If the user requested an audit only... If they already authorized refactoring") and a verification checkpoint ("Check material citations and reasoning with checklists/analysis-checklist.md"). Matches anchor 4 rather than 5 because the validation loop is thin — the checklist citation is the only explicit checkpoint, and the referenced checklist file does not exist in the bundle. | 4 / 5 |
Progressive Disclosure | The body is an overview with well-signaled, one-level-deep references tied to conditions ("Consult references/solid-ddd-context.md only for a domain-model question"), and a template asset is correctly externalized. Scored against the actual bundle: all referenced files exist except "checklists/analysis-checklist.md", which is dangling, and "scripts/validate-metadata.py" is present but unreferenced — minor organization gaps that hold this at anchor 4 instead of 5. | 4 / 5 |
Total | 17 / 20 Passed |