Content
100%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.
An exemplary terse reference skill: dense, accurate, well-sectioned, and free of padding. It delivers exactly what the description promises — name, semantics, and legality per decorator — with useful extras (constructor aliases, CLI overrides, doc anchors) at no token cost.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is a lean lookup table: every line carries semantics (exact names, node legality, constructor aliases like FDescribe/PDescribe/XIt, CLI overrides like --flake-attempts, constraints like 'No &|!,()/" chars'). It assumes Ginkgo knowledge and explains nothing Claude already knows, matching the anchor 'every token earns its place'. | 5 / 5 |
Actionability | For an instruction-only reference skill, the guidance is fully actionable: an inline usage example (It("...", Serial, Label("slow"), func(){...})), exact decorator signatures (FlakeAttempts(n), SpecPriority(int), AroundNode's three forms), and per-item semantics. Per the rubric's scoring note, absence of code is not penalized when the guidance is this concrete. | 5 / 5 |
Workflow Clarity | Single-purpose lookup skill with an unambiguous use pattern ('Terse lookup' up top, items grouped by purpose), so the simple-skill exception applies. No destructive or batch operations exist, so no validation checkpoints are required. | 5 / 5 |
Progressive Disclosure | The body is under 50 lines with no need for external reference files (none exist in the bundle), and is well-organized into clear sections by category, with a single one-level-deep pointer to the official docs — the rubric's under-50-lines guideline for a score of 5 is satisfied. | 5 / 5 |
Total | 20 / 20 Passed |