Content
85%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 a highly actionable, well-sequenced audit procedure with explicit verification checkpoints and concrete search/fix guidance throughout. Its main weaknesses are the absence of any progressive disclosure (everything is inline in a long file) and trimmable sections like the invocation examples.
Suggestions
Move the Phase 2 anti-pattern catalog and Phase 4 compound-findings table into reference files (e.g. references/anti-patterns.md, references/compound-findings.md), keeping SKILL.md as an overview with clearly signaled links
Cut the 'Invocation Examples' and 'Scope' sections — that trigger guidance belongs in the frontmatter description, not the body
Trim the 'Tool Use Is Mandatory' section to a single directive line, since its points repeat across phases
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Core sections are dense with operational content (grep patterns, fixes, tables) and assume Core Data knowledge, but the 'Invocation Examples' section, the 'Scope' section repeating the description, and repeated tool-use directives are padding that could be trimmed. | 4 / 5 |
Actionability | Provides exact grep strings ('NSMigratePersistentStoresAutomaticallyOption', 'relationshipKeyPathsForPrefetching'), copy-paste-ready Swift fix examples, a complete output template, a health-score rubric, and a false-positive list — fully executable guidance covering the common cases. | 5 / 5 |
Workflow Clarity | Five phases are explicitly sequenced with validation gates ('use Read to verify the surrounding context before reporting', 'Require evidence from the Phase 1 map', present the architecture map before proceeding) plus false-positive and output-limit checklists; the destructive-operation cap does not apply since this is a read-only audit. | 5 / 5 |
Progressive Disclosure | The ~280-line body is well-sectioned and its 'Related' pointers are clear one-level references, but the full anti-pattern catalog, compound-findings table, and invocation examples are all inlined with no reference files to split into, which a skill this long should use. | 3 / 5 |
Total | 17 / 20 Passed |