Content
81%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.
A concise, well-organized conceptual skill body that respects the reader's intelligence and ends with a clear stop condition. The main gap is actionability: it prescribes forms (end with the decision, options with costs) without a worked example or template that shows what a compliant report looks like.
Suggestions
Add one short worked example or before/after sketch showing a judgment-oriented report (e.g., a two-line before 'restates what was done' vs. an after that ends with a decision, options, costs, and a recommendation) to make the prescribed forms concrete.
Add brief section headers (e.g., 'Choosing the form', 'Reporting on work', 'Failure modes') so the reader can locate the guidance relevant to their situation without scanning the full prose.
Convert the reporting prescriptions ('give the options that matter, what each costs, and which you recommend') into a compact checklist format so the actionable sequence is scannable rather than embedded in a paragraph.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~40-line body is lean and dense: every paragraph carries skill-specific judgment guidance ("Show the governing relationship before its machinery", "Simplification fails when it hides uncertainty"), with no padding and no explanation of concepts Claude already knows. Every token earns its place. | 5 / 5 |
Actionability | Some concrete directives are present — "end with it, after the evidence it depends on", "give the options that matter, what each costs, and which you recommend", "Draw it from the work itself, such as the diff or the tests" — but the guidance stays conceptual throughout, with no worked example or before/after template demonstrating the form. Not level 4 because 'mostly executable' guidance would show concrete exemplars of the prescribed forms. | 3 / 5 |
Workflow Clarity | This is a simple, single-purpose skill under 50 lines with no multi-step process, so the simple-skill exception applies: the single action (shape the presentation around the reader's judgment) is unambiguous, and the body reads as a coherent progression ending in an explicit terminal checkpoint — "Stop when the person can make the next judgment without reconstructing the machinery behind it." No destructive or batch operations, so no validation cap applies. | 5 / 5 |
Progressive Disclosure | The body is short enough that no content needs splitting, and the one cross-link ([dialectic](../dialectic/SKILL.md)) is clearly signaled and one level deep; no nested references or inlined reference material. Not level 5 because the body has no section headers to aid navigation, and the referenced sibling-skill path cannot be verified within the bundle. | 4 / 5 |
Total | 17 / 20 Passed |