Claude Skills

A collection of 40 production-grade skills for Claude Code, designed for real-world applications.

Claude Skills

40 production-grade skills for Claude Code — deeply researched, battle-tested prompts with progressive disclosure architecture.

License: MIT Skills: 40 References: 200+


Why This Exists

Most skill collections give Claude generic instructions it already knows. This one is different:

  • Production patterns, not tutorials. Every skill encodes patterns from real systems — the kind of knowledge that comes from debugging at 3am, not reading docs.
  • Progressive disclosure. Core SKILL.md stays lean (~80 lines). Deep reference files load on-demand. Lower token cost, faster responses.
  • Opinionated. Instead of "here are 5 options", each skill recommends the best approach and explains why.
  • Anti-patterns included. Every skill explicitly calls out what goes wrong and why.

Quick Start

# Clone
git clone https://github.com/candeploys/claude-skills.git

# Copy all skills
cp -r claude-skills/skills/* ~/.claude/skills/

# Or cherry-pick
cp -r claude-skills/skills/react-mastery ~/.claude/skills/
cp -r claude-skills/skills/typescript-wizard ~/.claude/skills/

Architecture

Each skill follows a three-level progressive disclosure pattern:

skills/react-mastery/
├── SKILL.md                    # Core instructions (~80 lines)
│   ├── Role & expertise level
│   ├── Decision framework
│   ├── Core workflow steps
│   └── Reference routing table
└── references/                 # Loaded only when needed
    ├── hooks-patterns.md       # Custom hooks, optimization
    ├── state-management.md     # Zustand, Jotai, signals
    ├── server-components.md    # RSC architecture
    ├── performance.md          # Profiling, memoization
    └── testing.md              # Testing Library, MSW

How it works:

  1. Skill loads with minimal context (~80 lines)
  2. Claude reads the routing table to understand what references exist
  3. Claude loads only the specific reference files needed for the current task
  4. Result: 50%+ token reduction, surgical precision

All 40 Skills

Frontend (7)

SkillFocus
react-masteryReact 19, RSC, hooks composition, state, performance profiling
nextjs-architectApp Router, Server Actions, ISR/SSG, middleware, caching
vue-specialistComposition API, Pinia, Nuxt 3, reactivity system
svelte-expertRunes, SvelteKit, reactive declarations, stores
css-craftsmanModern CSS, Grid/Subgrid, container queries, animations, architecture
tailwind-masterDesign systems, custom plugins, responsive patterns, dark mode
graphql-expertSchema design, DataLoader, N+1, subscriptions, federation

Backend (7)

SkillFocus
nodejs-backendExpress/Fastify, streams, worker threads, clustering
python-expertAsync, type hints, packaging, generators, decorators
fastapi-proDependency injection, Pydantic v2, background tasks, WebSocket
django-masterORM mastery, DRF serializers, signals, custom managers
go-engineerGoroutines, channels, interfaces, error handling, testing
rust-systemsOwnership, lifetimes, async, traits, unsafe, FFI
api-designerREST maturity, versioning, pagination, rate limiting, HATEOAS

Data & Databases (4)

SkillFocus
sql-optimizerEXPLAIN plans, index strategies, window functions, CTEs
postgres-expertExtensions, partitioning, JSONB, full-text search, replication
database-designNormalization, schema evolution, migration strategies
redis-cachingData structures, pub/sub, Lua scripting, cluster mode

DevOps & Infrastructure (7)

SkillFocus
docker-orchestratorMulti-stage builds, Compose, security scanning, optimization
kubernetes-opsDeployments, HPA, RBAC, debugging, Helm charts
terraform-infraModules, state management, workspaces, providers
aws-architectService selection, IAM policies, cost optimization, Well-Architected
cicd-masterGitHub Actions, GitLab CI, matrix builds, caching
devops-pipelineRelease strategies, feature flags, canary, blue-green
nginx-configReverse proxy, SSL/TLS, load balancing, rate limiting

Architecture & Patterns (4)

SkillFocus
microservices-patternsSaga, CQRS, circuit breaker, service mesh, event sourcing
websocket-realtimeSocket.IO, scaling, presence, rooms, reconnection
auth-patternsOAuth2/OIDC, JWT, RBAC/ABAC, session management, MFA
monitoring-observabilityPrometheus, Grafana, OpenTelemetry, alerting, SLOs

Languages & Tools (5)

SkillFocus
typescript-wizardAdvanced generics, conditional types, template literals, decorators
git-workflowBranching strategies, interactive rebase, bisect, hooks
shell-scriptingBash best practices, error handling, process management
linux-adminSystemd, networking, permissions, troubleshooting, security
regex-masterPattern design, lookaheads/lookbehinds, performance, engines

Quality & Process (6)

SkillFocus
testing-arsenalUnit/integration/E2E strategies, mocking, coverage, TDD
security-hardeningOWASP Top 10, input validation, CSP, secrets management
performance-tunerProfiling, caching layers, lazy loading, CDN, Core Web Vitals
debugging-detectiveSystematic debugging, profiling, log analysis, bisection
code-reviewerReview checklists, refactoring patterns, PR best practices
documentation-writerAPI docs, ADRs, runbooks, READMEs, changelog

Usage

Skills activate automatically based on your request context:

"Build a React component with Server Components"
→ Activates: react-mastery
→ Loads: references/server-components.md

"My Postgres query is slow, help me optimize"
→ Activates: sql-optimizer + postgres-expert
→ Loads: references/query-plans.md, references/indexing.md

"Set up GitHub Actions CI/CD for my monorepo"
→ Activates: cicd-master + devops-pipeline
→ Loads: references/github-actions.md, references/monorepo.md

Contributing

See CONTRIBUTING.md for guidelines. The short version:

  1. Create skills/your-skill/SKILL.md with YAML frontmatter
  2. Add reference files in skills/your-skill/references/
  3. Keep SKILL.md under 100 lines, references under 300 lines each
  4. Include anti-patterns and "why" explanations
  5. Test with real prompts before submitting

License

MIT — See LICENSE