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 lean skill body: exact tool invocations, precise response fields, a defined output order, and explicit degraded-data and tool-refusal handling, all with zero wasted tokens. Nothing in the body fails to earn its place.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every line instructs: a three-row tool/access table, exact field names ("cii.combinedScore", "advisoryLevel", "sanctionsCount"), and terse rules. There is no explanation of concepts Claude already knows and no padding, matching the lean-and-efficient top anchor. | 5 / 5 |
Actionability | Guidance is fully executable in tool-call terms: named tools with exact parameters ("get_country_risk with country_code", "history_days: 30", "countries: [\"<code>\"]"), the exact response fields to read, and an explicit output section order. For an instruction-only skill this is copy-paste-ready, matching the top anchor. | 5 / 5 |
Workflow Clarity | Four clearly sequenced steps with explicit failure handling: "If upstreamUnavailable is true, say the risk snapshot is degraded" and "If a subscription tool is refused, deliver the sections that did load and name the ones that did not" provide error-recovery guidance for the realistic failure modes. The skill is read-only, so the destructive/batch validation cap does not apply, and this fits the explicit-validation top anchor better than the minor-gaps anchor. | 5 / 5 |
Progressive Disclosure | The body is under 50 lines, self-contained with no external references needed, and organized into clean Tools / Steps / Rules sections with no bundle files to misroute — matching the rubric's simple-skill provision that well-organized sections alone merit the top score. | 5 / 5 |
Total | 20 / 20 Passed |