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 an exemplary diagnostic workflow — ordered, prioritized, verification-driven, and packed with concrete, executable RevenueCat specifics. Its one structural flaw is progressive disclosure: the five referenced platform files are missing from the bundle, and substantial reference material is inlined rather than split out, so the carefully designed file split is not actually delivered.
Suggestions
Include the five referenced platform files (platforms/ios.md, platforms/android.md, platforms/kmp.md, platforms/flutter.md, platforms/react-native.md) in the bundle — the workflow currently dead-ends at 'read platforms/X.md' steps with no file to read.
Move the inline reference material (SDK error-code tables, known platform issues, and per-platform configuration checklists) into the corresponding platform files or a references/ file so SKILL.md remains a lean overview.
Deduplicate facts repeated across sections: the debug-logging setup in checklist item 1 vs. the 'debug log interpretation' reference section, the 'product_id:base_plan_id' format in Check 2 vs. the Android checklist, and the INVALID_CREDENTIALS explanation in checklist item 2 vs. the error-code table.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with RevenueCat-specific facts Claude cannot reliably know (API key prefixes, list-offerings pagination limits, iOS 18.x StoreKit bugs, Play testing-track requirements) with no generic concept explanations, but several facts are duplicated across sections — debug-logging setup appears in checklist item 1 and the 'debug log interpretation' reference, the 'product_id:base_plan_id' format appears in Check 2 and the Android checklist, and INVALID_CREDENTIALS is explained twice — leaving minor trimmable redundancy. | 4 / 5 |
Actionability | Guidance is fully executable: named MCP tools with exact parameters ('list-offerings' with 'limit: 100' and 'expand=items.package.product'), rc CLI command mappings, copy-paste Swift/Kotlin snippets for cache-busting getCustomerInfo, exact logLevel settings, specific error codes, and precise settings paths for sandbox accounts. | 5 / 5 |
Workflow Clarity | A clearly sequenced five-stage workflow (detect platform → nine-item ordered checklist → three-phase dashboard inspection → platform specifics → verify the fix) with explicit prioritization ('Most reports are resolved by steps 1 through 5'), a dedicated verification section with three checkpoints, a feedback loop (request full debug logs if reproduction fails), and user confirmation before executing fixes. | 5 / 5 |
Progressive Disclosure | The body instructs 'read platforms/ios.md' (and android, kmp, flutter, react-native) at multiple points, but none of these files exist in the bundle — no platforms/, references/, scripts/, or assets/ directories are present — so navigation dead-ends exactly where the skill delegates detail, and ~100 lines of reference tables (error codes, known platform issues, per-platform checklists) remain inlined in SKILL.md. | 3 / 5 |
Total | 17 / 20 Passed |