Conciseness

The body is compact (~35 lines), but there is noticeable redundancy across sections: BCP 47 validity is stated in Quick Reference, Check, Fix, and Explain; empty `lang=''` is flagged in both Quick Reference and Check; and the Explain section re-describes BCP 47 subtag structure (primary vs. region subtags) that Claude already knows. Not a 2 because nothing is padded filler and each section has a distinct purpose (Check/Fix/Explain).

Actionability

The Check section gives a concrete numbered verification ("(1) a `lang` attribute is present; (2) its value is non-empty; (3) its value is a valid BCP 47 language code") plus an explicit flag list ("missing `lang`, empty `lang=''`, language codes using full names ... mismatches"), and Fix provides copy-paste-ready snippets like `<html lang='fr'>`. Not a 5 because the body itself has no executable code example (correct/incorrect HTML lives in references/rule.md) and no command for validating against the IANA registry.

Workflow Clarity

This is a simple, single-purpose skill and the action is unambiguous: a numbered inspect-and-verify sequence in Check, explicit flag conditions, a Fix section with exact remediation per case, and a validation directive ("Validate language codes against the IANA Language Subtag Registry"). The simple-skill exception applies; no destructive or batch operation requires a feedback loop.

Progressive Disclosure

The SKILL.md body is a well-organized overview under 50 lines with clear sections (Quick Reference / Check / Fix / Explain / Code Review), and the one bundle file `references/rule.md` exists and is clearly signaled at the end ("For full implementation details, code examples, and framework-specific guidance, see `references/rule.md`"). References are one level deep with no nested chaining.

Specificity

Quotes like "Check the `<html>` opening tag for a `lang` attribute with a non-empty, valid BCP 47 language code" and "check for `lang` attribute changes on individual elements when the document contains content in multiple languages" list several concrete actions with precise technical criteria. Not a 5 because the closing sentence about inheritance is informational rather than an action, and the "Use when applies to all HTML documents" clause is garbled.

Completeness

It answers both: 'what' ("Check the `<html>` opening tag for a `lang` attribute with a non-empty, valid BCP 47 language code" plus checking per-element lang changes) and 'when' ("Use when applies to all HTML documents"). Not a 5 because the 'when' clause is ungrammatical and maximally broad rather than giving concrete trigger phrases; not a 3 because an explicit 'Use when' trigger is clearly present.

Trigger Term Quality

Relevant keywords are present ("HTML documents", "lang attribute", "BCP 47", "multilingual"), but common natural variations a user would say are missing: "accessibility", "screen readers", "language tag", "i18n", "locale". Better than a 2 because the core term users would naturally use for this exact task (`lang` attribute) is present and repeated.

Distinctiveness Conflict Risk

The niche is well defined — validating the `lang` attribute and BCP 47 codes — so it would rarely trigger for unrelated skills. Minor overlap risk with general HTML/accessibility skills because "applies to all HTML documents" makes the trigger extremely broad rather than scoped to language/accessibility concerns.