Communitygithub.com

yagyaraj234/skills

Check frontend readiness from a working diff, PR URL, release target, or full repository audit. Use for app, UI, accessibility, performance, and frontend PR reviews.

What is skills?

skills is a Claude Code agent skill that check frontend readiness from a working diff, PR URL, release target, or full repository audit. Use for app, UI, accessibility, performance, and frontend PR reviews.

Works with✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/yagyaraj234/skills/tree/HEAD/skills/frontend-readiness

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Frontend readiness

Invoke as $frontend-readiness diff, $frontend-readiness pr <URL>, $frontend-readiness release, or $frontend-readiness full.

Use the mode, evidence, verdict, and edit rules in check-readiness. When invoked directly, select a mode there. When invoked by check-readiness, keep its selected target. Assess frontend surfaces only.

Assess checks that apply to the chosen target. Mark an in-scope exception N/A with a reason. A section that does not apply (non-public app for Search and sharing, no framework for Framework conventions) is one N/A line, not one per check.

Blocking areas: broken critical flow or route, auth and permission UI, unsafe rendering of user content, leaked secrets, accessibility failures on critical flows, missing error or empty states on data-driven surfaces, failing build or tests. Other failures are advisory.

Correctness

  • Every in-scope route renders through direct navigation, refresh, in-app navigation, back, and forward.
  • Every critical flow completes its happy path and preserves useful input on recoverable failure.
  • Forms validate input, prevent duplicate submission, show actionable errors, and expose errors to assistive technology.
  • Client, server, URL, and persisted state remain consistent across refresh, cancellation, stale responses, and repeated actions.
  • Console has no new errors or warnings; failed network requests are explained.
  • Auth and permission boundaries show correct UI, preserve intended return routes, and handle expired sessions, forbidden responses, and role changes.

Every data-driven surface has an explicit state matrix. Mark a state N/A only when it cannot occur or offers no user-facing value.

  • Empty: successful zero-data result explains what happened and offers a useful next action.
  • Loading: progress is clear, useful existing content stays stable, and repeated actions stay controlled.
  • Skeleton: where final layout is predictable and delay is perceptible, placeholders match it and avoid layout shift; otherwise record N/A.
  • Partial: available content stays usable while independent sections load or fail.
  • Error: message is actionable, useful input is preserved, and retry or recovery works.

Framework conventions

Skip items for frameworks the project does not use; name the detected framework and version.

  • Framework, router, and data-loading APIs match the installed version and existing project conventions; record the version and source of truth.
  • Internal navigation uses the detected framework link component, such as Link from next/link or @tanstack/react-router; native <a> handles external, download, email, and phone links.
  • Route params, search params, redirects, and navigation use the framework router APIs and hooks.
  • Static and dynamic styles follow the existing framework-supported styling system; dynamic values use CSS custom properties unless library-required positioning needs inline styles.

Interface and accessibility

  • Layout works at each recorded viewport without unintended overflow. Long text, missing media, slow data, and large collections remain usable.
  • Hierarchy, spacing, typography, colors, and interactive feedback are consistent; controls cover hover, focus, active, disabled, loading, and error states.
  • Destructive actions require clear intent and offer recovery where practical.
  • Semantic landmarks, headings, controls, labels, and accessible names match purpose.
  • Keyboard-only use reaches and operates every interactive element in logical order; focus stays visible and moves correctly through dialogs, menus, errors, and route changes.
  • Text and essential UI meet the recorded accessibility target (WCAG 2.2 AA unless a stricter requirement applies); meaning does not depend on color alone.
  • Images have appropriate alternatives; decorative media stays silent. Zoom, text resizing, reduced motion, and required screen-reader announcements preserve the flow.

Resilience, privacy, and security

  • User-controlled content is rendered safely; secrets and privileged decisions remain server-side.
  • External links, redirects, uploads, storage, and browser APIs validate untrusted input. External tabs use safe opener behavior where applicable.
  • Offline and timeout behavior avoids data loss.
  • Error UI omits sensitive internals while logs retain enough diagnostic context.
  • Authentication or personal-data products expose public Terms and Privacy pages from auth and persistent navigation; content matches actual behavior.
  • Cookies, CSRF defenses, response security headers, redirects, and cache policy are verified when the application controls them.
  • Analytics, error tracking, and consent respect privacy requirements and do not expose personal or secret data.
  • Feature flags have a verified target-environment value and, for risky changes, a documented rollback or disable path.

Performance and assets

  • Representative routes are measured from a production build in the recorded environment using TTFB, LCP, INP, CLS, first useful content, and primary-interaction timing.
  • Every in-scope route shows useful content or an actionable fallback and accepts its primary interaction within the recorded budget; five seconds is the default only when no stricter product budget exists.
  • Production build output, route chunks, and network waterfall contain no unexplained large cost.
  • Changed performance optimizations have measured benefit and preserve keyboard, focus, screen-reader, and offline behavior where applicable.
  • Effects release listeners, subscriptions, observers, pending requests, workers, timers, animation frames, and debounced or throttled work when dependencies change or components unmount.
  • Images use available framework optimization, responsive dimensions and sizes, modern formats, compression, stable aspect ratios, and correct eager/lazy priority. Video and fonts have deliberate delivery, fallback, and preload behavior.
  • Third-party, inline, and blocking scripts have an explicit purpose and loading priority.

Search and sharing

Applies only to public or shareable routes; authenticated-only apps need just the non-indexing check.

  • Every route defines appropriate title and description metadata. Public indexable routes set canonical and robots intent through framework APIs; non-production and protected environments are not accidentally indexed.
  • Shareable routes resolve Open Graph and social metadata to absolute, crawlable values. OG images return the correct content type, dimensions, and alt text; rendered tags and a real share preview are verified.
  • For a release gate, Lighthouse runs against deployed production routes on recorded mobile and desktop profiles; Performance, Accessibility, Best Practices, and SEO results and regressions are recorded.

Engineering and deployment

  • Formatter, lint, typecheck, tests, and production build pass or have an explicit baseline failure.
  • Changed logic has the smallest useful automated regression check.
  • Environment variables, feature flags, API URLs, browser support, and test accounts match the target environment.
  • Error tracking, favicons, and install icons work where the product requires them.
  • For a release gate or deployment-affecting change, deployed assets, routes, APIs, auth, headers, redirects, and cache behavior are verified in the target environment.

Individual skills in this repo

This repo contains 1 individual skill — each has its own dedicated page.

Related Skills