Content
92%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 tightly written, highly actionable advisory skill with a clear sequenced workflow and excellent references/ routing. The one real defect is that two workflow-critical checklist files (decision-tree.md, ship-readiness.md) are referenced but absent from the bundle, which breaks progressive disclosure for the skill's own gating steps.
Suggestions
Add the missing checklists/decision-tree.md and checklists/ship-readiness.md files (or remove the references and inline the gating criteria), since both the workflow and the output-style section depend on them existing.
If the checklists are intentionally deferred, replace the bare file references with a brief inline description of what each gate checks so the skill remains usable when the files are absent.
Double-check every path mentioned in the body (e.g., references/03-webview-survival.md § "Hidden window throttling") resolves to a real file and section, to keep navigation one-level-deep and intact.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~41-line body is dense and high-signal with no padding of concepts Claude already knows — it jumps straight into tenets, routing, anti-patterns, and trade-offs. It is not level 2 because there is no unnecessary explanation; even the long "one-paragraph version" is a deliberate compression that earns its place as orientation. | 3 / 3 |
Actionability | For an advisory skill it gives concrete, specific guidance: a question→reference routing table, per-anti-pattern recommendations pointing at exact files ("Recommend references/02-architecture.md instead"), and concrete architectural specifics (Swift/AppKit, C#/WPF, Rust+UniFFI, ~400MB). It is not level 2 because the guidance is complete and specific rather than abstract or pseudocode-like, and the scoring note explicitly does not penalize absence of code in instruction-only skills. | 3 / 3 |
Workflow Clarity | "How to use this skill" is a clearly numbered 4-step sequence with explicit gating checkpoints ("Before recommending... run the decision tree", "Before the user claims... run ship-readiness"). It is not level 2 because validation checkpoints are explicit and named, not implicit. This is an advisory (not destructive/batch) skill, so the missing-feedback-loop cap does not apply. | 3 / 3 |
Progressive Disclosure | The references/ layer is excellent — all 7 referenced files exist, one level deep, with a clear routing guide. However the workflow's two gating checkpoints point to checklists/decision-tree.md and checklists/ship-readiness.md, and no checklists/ directory exists in the bundle, so those references are broken. Scored against the actual bundle structure (per guideline), this navigation defect keeps it below 3; it is not level 1 because the references that do exist are well-organized and clearly signaled rather than monolithic. | 2 / 3 |
Total | 11 / 12 Passed |