Content
100%Weight 40%Scale 1-3Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured, lean instruction skill: it sequences a multi-step research workflow with explicit validation gates, hands Claude concrete tool fields and a rendering snippet, and offloads craft details to real, clearly signaled one-level-deep references. It does not re-explain concepts Claude already knows.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and point-first — it assumes Claude knows what IRAC, a memo, and citation are and never pads with such concepts; every line is task-specific imperative guidance, so it sits at the lean/efficient anchor rather than the 'could be tightened' one. | 3 / 3 |
Actionability | It gives concrete, executable guidance: named Midpage tools with exact filter params (jurisdictionType, circuits/courts/states, publishStatus), specific result fields (doesNotAddress, supportedPropositions, deeplinkURL, centrality, scope, opinionSection), and a copy-paste rendering snippet using D.builders("memo", { lineSpacing: 480 }). | 3 / 3 |
Workflow Clarity | The research method is a clearly sequenced pipeline (Frame → Search → Triage → Branch → analyzeOpinion gate) with explicit validation checkpoints — "Check doesNotAddress first — if your point is listed, the case doesn't stand for it" — and a saturation stopping condition, matching the anchor for clear sequence with explicit validation steps. | 3 / 3 |
Progressive Disclosure | The body is an overview that signals three one-level-deep reference files at the top (litigation-writing.md, citations.md, legal-docx.md), all of which exist on disk, while keeping the core method inline — clear navigation with content appropriately split. | 3 / 3 |
Total | 12 / 12 Passed |