Lean, opinionated prose that assumes Svelte competence; every numbered principle earns its place with a tight one-line rule plus a concrete do/don't example, with no padding explaining what Svelte or runes are.

Provides fully executable, copy-paste-ready Svelte 5 snippets (e.g. `$derived(count * 2)`, `let { onSelect } = $props()`) with explicit bad/good pairs, plus a concrete review procedure and a smell-index table mapping each smell to a reference file.

The 'Running a Review' section gives a clear 4-step sequenced procedure including a validation checkpoint ('Detect the version first') that gates which syntax set to apply, with feedback logic for version mismatches; the smell index provides a checkable scan list.

SKILL.md is a concise overview with 20 one-level-deep, clearly signaled references (all verified to exist as real files in references/), each linked inline from its principle and aggregated in a navigable smell-index table; detail is appropriately split out rather than inlined.

Names concrete review actions ('Reviews Svelte components', 'identifies anti-patterns', 'suggests Svelte 5 refactorings') and enumerates many specific smell categories, matching the 'multiple specific concrete actions' anchor.

Explicitly answers both what (reviews components, identifies anti-patterns, suggests refactorings) and when via a clear 'Use when...' clause with concrete triggering situations, satisfying the top anchor.

Lists natural trigger phrases users would actually say ('is this idiomatic', 'Svelte way', 'doing Svelte 5 migration') plus concrete code-smell terms (createEventDispatcher, export let, on:click, clickable div), giving broad natural coverage.

Has a clear, narrow niche (Svelte 5 component idiomaticity) with distinct triggers and explicit scope boundaries ('delegates to sveltekit for routes/load/actions/server concerns'), making conflicts with other skills unlikely.