Content
65%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 code-rich, highly actionable reference for Biome diagnostics that mostly respects Claude's competence, but it is monolithic, repeats some material across the manual-advice and RuleDiagnostic sections, and lacks explicit validation checkpoints in its workflow.
Suggestions
Add an explicit verification step to the workflow, e.g. after registering a category and wiring the diagnostic, run `cargo test` or `biome check` on a sample to confirm it renders.
Collapse the overlap between 'Implement Advices' (custom advice type) and 'Add Diagnostic to Rule' (RuleDiagnostic builder) by leading with the RuleDiagnostic pattern most rules actually use and demoting the manual approach.
Split the type/API reference and the good/bad message gallery into a reference file under references/ so SKILL.md stays a lean overview, improving progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is code-forward and assumes Biome/Rust knowledge without explaining basics, but the manual-advice-types section overlaps with the later 'RuleDiagnostic builder pattern' note, and the 'Common Patterns' section reiterates patterns already shown above, so it could be tightened. | 2 / 3 |
Actionability | It provides fully executable Rust across derive macros, Advices impls, RuleDiagnostic::new().note(), markup! usage, and define_categories! registration, with real type names and field annotations that are copy-paste ready. | 3 / 3 |
Workflow Clarity | The 'Common Workflows' section sequences diagnostic creation steps clearly, but there are no explicit validation or verification checkpoints (e.g., running cargo test or biome check to confirm the diagnostic renders), leaving feedback loops implicit. | 2 / 3 |
Progressive Disclosure | No bundle files exist; the ~375-line SKILL.md is monolithic with clear headers and a one-level external References list, but content that could be split (type reference, good/bad message gallery) is inline rather than progressively disclosed into reference files. | 2 / 3 |
Total | 9 / 12 Passed |