skills.rest
civitai
civitaiOFFICIAL
Explore all skills by this author →
Website𝕏@HelloCivitaiGitHub

svelte-review

Runs three SvelteKit review agents over a feature segment and consolidates their findings.

7.2k|734|Updated Oct 11, 2022
Search Tags:#sveltekit#svelte 5#svelte-check#review agents#pre-commit review#Code review#retool migration
One-click install
npx skills add https://github.com/civitai/civitai --skill svelte-review
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: svelte-review
Source: https://github.com/civitai/civitai/tree/main/.claude/skills/svelte-review
Command: npx skills add https://github.com/civitai/civitai --skill svelte-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Reviewing SvelteKit feature work by hand is inconsistent: logic bugs, Svelte 5 idiom violations, and missing abstractions each need a different lens, and ported segments can faithfully implement the wrong behavior without anyone noticing. This Skill standardizes a pre-completion review so a segment is not called done until all angles are checked.

Core Features & Use Cases

  • Three parallel review lanes: Coordinates correctness (logic, data shape, auth scope, failure paths), Svelte 5 idiom plus shared UI conventions, and abstraction (duplication, missing components, placement) agents over one scoped segment.
  • Fidelity pass for ported code: Adds a fourth source-vs-implementation comparison when a segment was ported from Retool or the main Next.js app, catching absent capabilities the code-only reviews miss.
  • Consolidated ranked findings: Deduplicates overlapping reports, drops disproven findings, ranks by user-facing consequence, and separates fix-now items from deferred notes.
  • Use Case: Before committing a new moderator panel in apps/moderator, run this review to verify the diff against the Svelte app standard, then close out with svelte-check and a manual browser look at the page.

Quick Start

Review the current uncommitted segment in apps/moderator with the svelte-review process and list the consolidated findings before I commit.

Frequently Asked Questions about svelte-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I review a SvelteKit feature segment before committing?▼

Scope the segment with git diff against main, then run the three review agents covering correctness, Svelte 5 idiom, and abstraction over that file list. Consolidate their findings into one ranked list, run svelte-check, and inspect the page in a browser before committing.

What does a Svelte 5 code review check for?▼

It checks logic and data shape correctness, auth scope, failure paths, Svelte 5 idiomatic patterns, shared UI and styling conventions, and duplication or missing component abstractions. Each area is covered by a dedicated review agent with the app directory as scope.

How do I verify code ported from Retool matches the original?▼

Run a dedicated fidelity pass that walks the Retool export source item by item and confirms each behavior exists and matches in the built files. Code-only reviews compare the code to itself and will pass over a faithful implementation of the wrong thing.

Should Svelte review agents run in parallel or serially?▼

Serial execution is the safe default because running three agents at once is heavy and has coincided with sessions dying mid-review. Spawn them together only when the segment under review is small.

Why does svelte-check pass but the page still render blank?▼

Typecheck only validates types and cannot detect rendering or runtime behavior bugs, so pages can pass while displaying nothing. Always open the page in a browser after typecheck as the final verification step.

Related Skills

Explore More Software Engineering Skills→
mia-cxmia-cx

web-app-builder

Scaffold and deploy SvelteKit multi-page web apps to Cloudflare Pages.

Official
Advanced
swyxioswyxio
154

autoreview

Run structured closeout code reviews on branches, PRs, and commits.

Community
Advanced
DataDogDataDog
648

lint

Run Hatch-based lint, format, and quality checks on modified files.

Official
Advanced
aslaiiaslaii

gsd-fast

Execute trivial tasks directly without spawning subagents or creating PLAN.md files.

Community
Basic