Content
80%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 body is an exemplary lean detection guide: a single memorable decision rule, a concrete anti-pattern checklist, explicit do-not-flag guardrails against false positives, and a correct-pattern snippet. Its main weaknesses are the missing validation loop for the default-on, potentially batch fix mode and thin guidance on what applied fixes should look like.
Suggestions
Add a validation checkpoint to the Steps, e.g. "After applying fixes, run the typecheck/tests for the scope and grep for any remaining references to the removed callback before finishing" — this would lift workflow_clarity past the batch-operation cap of 3.
Show one concrete before/after fix example (e.g. unwrapping an unobserved useCallback) so the fix step is as executable as the detection step.
Merge anti-patterns 1 and 3, since native-element handlers are already covered by pattern 1's "no observer" case, removing the only duplicated content.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence: it never explains what useCallback or React.memo are, jumps straight to the decision rule ("useCallback is only useful when something observes the reference"), a terse observer checklist, seven named anti-patterns, and a single compact correct-pattern snippet. Every section earns its place; the only near-redundancy (native element handlers appear in patterns 1 and 3) is minor and does not pad the text. Matches anchor 5 ("Lean and efficient; every token earns its place"). | 5 / 5 |
Actionability | Detection guidance is concrete and executable: named patterns with code-shaped examples ("<button onClick={fn}>", "a dep is a plain object/array created inline"), an explicit do-not-flag list, and a copy-paste-ready correct ref-pattern snippet. It falls short of anchor 5 because the fix side is underspecified — step 3 says only "apply the fixes" without showing what a fix looks like (e.g. the before/after of unwrapping a useCallback or moving to useMemo). | 4 / 5 |
Workflow Clarity | The three steps (read reference, analyze scope, apply or propose fixes) form a clear sequence, but the skill defaults to fix=true and its scope can be "whole codebase" — a batch code-modification workflow with no validation checkpoint (e.g. re-check for remaining observers, run typecheck/tests) before or after applying fixes. Per the guidelines, batch operations without validation cap workflow clarity at 3; the cap also matches anchor 3 ("sequence present but checkpoints missing"). | 3 / 5 |
Progressive Disclosure | The skill is a single-purpose, ~54-line body with no bundle files (no references/, scripts/, or assets/ directories exist), and it is organized into well-labeled sections (References, the rule, anti-patterns, correct patterns, Steps) with its one external reference (react.dev) clearly signaled under a References heading. Per the scoring notes, a sub-50-line skill with no need for external references scores 5 with well-organized sections, which this matches. | 5 / 5 |
Total | 17 / 20 Passed |