Content
77%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 well-structured, appropriately split rule skill with an unambiguous check/fix workflow and a correctly used one-level reference. The main weakness is redundancy — the identical-values rule and the HTML5-only-`lang` point are restated across four sections, and the Explain section teaches concepts Claude already knows.
Suggestions
Consolidate the repeated rule statements: state 'values must be identical, including subtags' once in Quick Reference and let Check/Fix reference it instead of restating it.
Trim the Explain section to the non-obvious part (why polyglot documents need both attributes) and drop the restatement of what `lang` and `xml:lang` do for HTML vs XML parsers.
Move the generic Code Review paragraph into references/rule.md (where verification tooling like axe/Lighthouse is already listed) or name the specific tools inline to remove the vague 'browser accessibility tooling' phrasing.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is short and organized, but the core rule is repeated across the intro paragraph, Quick Reference, Check, Fix, and Explain sections (e.g., 'their values must be identical' appears in three places, and the HTML5-only-needs-`lang` point appears four times), and the Explain section restates well-known concepts ('The `lang` attribute on `<html>` serves this purpose for HTML parsers'). This fits anchor 3, 'mostly efficient but includes some unnecessary explanation or could be tightened' — above 2 because there is no padded tutorial-style filler, below 4 because the repetition and the known-concept explanation are more than minor. | 3 / 5 |
Actionability | The Check and Fix sections give concrete, executable instruction: 'Inspect the `<html>` element for both `lang` and `xml:lang` attributes', 'including subtags — `en` vs `en-US` is a mismatch', and document-type-specific fixes ('remove `xml:lang`' for text/html, 'keep both and ensure they are identical' for XHTML/polyglot). It falls short of 5 because there are no inline code/markup examples (correct/incorrect `<html>` tags live only in references/rule.md) and the Code Review section is vague ('browser accessibility tooling or assistive tech' with no named tool); it is above 3 because the guidance is specific and covers the common cases. | 4 / 5 |
Workflow Clarity | This is a simple, single-purpose skill under 50 lines whose single action is unambiguous: the Check → Fix → Explain → Code Review sections cleanly sequence detect, correct, justify, and review, with the decision branches (HTML5 vs XHTML vs polyglot) each handled explicitly. Per the rubric's simple-skill exception the single action scores 5; no destructive/batch validation cap applies since this is an audit plus a one-attribute fix. | 5 / 5 |
Progressive Disclosure | The body is a lean overview with clear section headers and a single, clearly signaled one-level-deep reference — 'For full implementation details, code examples, and framework-specific guidance, see `references/rule.md`' — and that file exists in the bundle, holding exactly the deeper material (code examples, per-document-type table, standards links, verification steps). This matches anchor 5: clear overview, well-signaled reference, content appropriately split. | 5 / 5 |
Total | 17 / 20 Passed |