Content
81%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 delivers an exceptionally actionable, well-sequenced closed-loop debugging workflow with explicit verification and escalation rules — its strongest qualities. The main weaknesses are redundancy (the failure-pattern catalogue appears three times in different forms) and a monolithic single-file structure where reference material is inlined. Trimming the duplicated tables and splitting lookup content into reference files would improve both.
Suggestions
Consolidate the three redundant presentations of the failure-pattern mapping (Failure Pattern Recognition table, Decision Tree branch, Error Quick Reference table) into one authoritative table referenced by the others.
Move the stable lookup content — the xcsym pattern_tag table and the failure-pattern/fix catalogue — into a references/ file, keeping the workflow phases lean in SKILL.md.
Trim the Invocation Examples, Related, and Scope sections at the end, which restate routing information already covered by the description and decision tree.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient — exact commands carry high-value, non-obvious commentary (e.g. never pipe xcodebuild through tee; shell variables don't persist between tool calls) — but the pattern-to-fix mapping is repeated three times (Failure Pattern Recognition table, Decision Tree, Error Quick Reference table) and the Invocation Examples/Related sections pad the tail. Not 4 because triple redundancy exceeds 'minor instances of over-explanation'. | 3 / 5 |
Actionability | Fully executable, copy-paste-ready guidance: exact xcodebuild/xcresulttool/xcsym invocations with exit-code semantics, mktemp-based result-bundle handling, and concrete BEFORE/AFTER Swift fix examples covering the common failure patterns. Placeholders like <SCHEME_NAME>UITests are explained, so there are no gaps. | 5 / 5 |
Workflow Clarity | The phased RUN→CAPTURE→ANALYZE→SUGGEST→FIX→VERIFY→REPORT loop has an explicit feedback path (re-run if still failing), per-fix verification (Phase 6), one-fix-at-a-time discipline, max-3-iteration escalation, and a warning before the destructive rm -rf of .xcresult bundles. Validation checkpoints are explicit, so the destructive-operation cap does not apply. | 5 / 5 |
Progressive Disclosure | A well-sectioned single file with clear headers and clearly signaled one-level-deep cross-references to sibling skills (axiom-testing, axiom-concurrency, xcsym-ref.md); no nested references. Not 5 because at ~360 lines the pattern catalogue, fix recipes, and invocation examples are all inlined — content that could be split into reference files for a skill this size. | 4 / 5 |
Total | 17 / 20 Passed |