Claude Skills

Comprehensive prompt engineering with Chain-of-Thought, few-shot learning, prompt versioning, and optimization. Use when designing prompts, improving accuracy, managing prompt lifecycle.

Testing patterns for LLM-based applications. Use when testing AI/ML integrations, mocking LLM responses, testing async timeouts, or validating structured outputs from LLMs.

Security patterns for LLM integrations including prompt injection defense and hallucination prevention. Use when implementing context separation, validating LLM outputs, or protecting against prompt injection attacks.

High-performance LLM inference with vLLM, quantization (AWQ, GPTQ, FP8), speculative decoding, and edge deployment. Use when optimizing inference latency, throughput, or memory.

Data visualization with Recharts 3.x including responsive charts, custom tooltips, animations, and accessibility for React applications. Use when building charts or dashboards with Recharts.

Progressive Web App patterns with Workbox 7.x, service worker lifecycle, offline-first strategies, and installability. Use when building PWAs, service workers, or offline support.

Code splitting and lazy loading with React.lazy, Suspense, route-based splitting, intersection observer, and preload strategies for optimal bundle performance. Use when implementing lazy loading or preloading.

Image optimization with Next.js 16 Image, AVIF/WebP formats, blur placeholders, responsive sizes, and CDN loaders. Use when improving image performance, responsive sizing, or Next.js image pipelines.

Dashboard UI patterns with widget composition, real-time data updates, responsive grid layouts, and data tables for React applications. Use when building dashboards, widgets, or data tables.

Core Web Vitals optimization for LCP, INP, CLS with 2026 thresholds, performance budgets, and RUM. Use when improving page performance, diagnosing CWV regressions, or setting performance budgets.

shadcn/ui component patterns including CVA variants, OKLCH theming, cn() utility, and composition. Use when adding shadcn components, building variant systems, or customizing themes.

HyDE (Hypothetical Document Embeddings) for improved semantic retrieval. Use when queries don't match document vocabulary, retrieval quality is poor, or implementing advanced RAG patterns.

Radix UI unstyled accessible primitives for dialogs, popovers, dropdowns, and more. Use when building custom accessible components, understanding shadcn internals, or needing polymorphic composition.

User personas, customer journey maps, empathy maps, and experience mapping patterns. Use when synthesizing research into actionable models, understanding user experiences, or aligning teams on customer needs.

Use this skill when creating or evolving design systems for applications. Provides design token structures, component architecture patterns, documentation templates, and accessibility guidelines. Ensures consistent, scalable, and accessible UI design across products.

End-to-end type safety with Zod, tRPC, Prisma, and TypeScript 5.7+ patterns. Use when creating Zod schemas, setting up tRPC, validating input, implementing exhaustive switch statements, branded types, or type checking with ty.

Integration testing patterns for APIs and components. Use when testing component interactions, API endpoints with test databases, or service layer integration.

React Aria (Adobe) accessible component patterns for building WCAG-compliant interactive UI with hooks. Use when implementing buttons, dialogs, comboboxes, menus, and other accessible components in React applications.

React Hook Form v7 with Zod validation, React 19 useActionState, Server Actions, field arrays, and async validation. Use when building complex forms, validation flows, or server action forms.

Transactional outbox pattern for reliable event publishing. Use when implementing atomic writes with event delivery, ensuring exactly-once semantics, or building event-driven microservices.

Message queue patterns with RabbitMQ, Redis Streams, and Kafka. Use when implementing async communication, pub/sub systems, event-driven microservices, or reliable message delivery.

Biome 2.0+ linting and formatting for fast, unified code quality. Includes type inference, ESLint migration, CI integration, and 421 lint rules. Use when migrating from ESLint/Prettier or setting up new projects.

Anthropic's Contextual Retrieval technique for improved RAG. Use when chunks lose context during retrieval, implementing hybrid BM25+vector search, or reducing retrieval failures.

Enforce 2026 folder structure standards - feature-based organization, max nesting depth, unidirectional imports. Blocks structural violations. Use when creating files or reviewing project architecture.