Content
77%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 well organized: a validated workflow with a fix-and-retry loop, a clean one-level-deep reference table backed by real files, and concrete code. Its main inefficiency is spending significant space on basic Kotlin (null-safety operators, scope-function semantics) and a filler technology-name list that Claude already knows, which is the primary scoring drag.
Suggestions
Trim the Null Safety and Scope Functions sections to bare constraint bullets (or move them into references/dsl-idioms.md) — they teach basic Kotlin operators and scope-function semantics Claude already knows.
Delete the "Knowledge Reference" paragraph; a bare list of 16 technology names adds no actionable guidance.
Remove or isolate the "Kotlin 1.9+" version pin (it appears twice) into a dedicated version/old-patterns note so time-sensitive details don't drift.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The workflow, reference table, and constraints are lean, but roughly a third of the body teaches basic Kotlin Claude already knows — the Null Safety section ("val displayName = user?.profile?.name ?: \"Anonymous\""), the Scope Functions section with explanatory comments ("apply — configure an object, returns receiver"), and the filler "Knowledge Reference" paragraph listing 16 technology names. The repeated time-sensitive "Kotlin 1.9+" version pin (not in a deprecated/old-patterns section) also penalizes per the guidelines. This is more than the 'minor instances' of anchor 4, fitting anchor 3's 'some unnecessary explanation or could be tightened'. | 3 / 5 |
Actionability | Guidance is mostly executable: complete, compilable Kotlin classes (sealed UiState hierarchy, UserRepository with flow builder and StateFlow), concrete tool commands ("Run detekt and ktlint", "runTest, Turbine"), and a 4-item output template. Minor gaps keep it below anchor 5: snippets call undefined helpers (showSpinner(), showUser(), api.fetchUser), and workflow steps like "Apply inline classes, sequence operations" are directives rather than runnable commands. | 4 / 5 |
Workflow Clarity | The 6-step Core Workflow is clearly sequenced with an explicit validation gate and feedback loop — "*If detekt/ktlint fails:* Fix all reported issues and re-run both tools before proceeding to step 5" — plus an inline checkpoint in step 3 ("Verify coroutine cancellation is handled... and null safety is enforced") and the MUST DO / MUST NOT constraint lists acting as checklists. This matches the anchor-5 pattern of explicit validation, error-recovery loops, and checklists. | 5 / 5 |
Progressive Disclosure | The body is a genuine overview: a Reference Guide table maps five topics to real files with clear "Load When" conditions, and all five referenced paths (references/coroutines-flow.md, multiplatform-kmp.md, android-compose.md, ktor-server.md, dsl-idioms.md) exist in the bundle. References are one level deep, clearly signaled, with only concise key-pattern examples kept inline — matching the anchor-5 clear-overview structure. | 5 / 5 |
Total | 17 / 20 Passed |