Skip to content

Best Skills for Backend and DevOps

The AI just generated an Express route handler with no error handling, a database query with no connection pooling, and an async function that swallows exceptions with an empty catch block. The code works in development. In production, it will leak connections, crash silently, and leave you debugging at 3 AM.

Backend skills teach the AI production patterns: proper error handling, connection management, security defaults, and the operational concerns that separate hobby code from production code. Install the right skills and the AI starts generating code that is ready for real traffic.

  • The top real skills for backend development, databases, and infrastructure
  • Skills for security auditing, authentication, and payment integration
  • What each skill teaches and why it matters for production
  • Strategies for combining backend skills with MCP servers

The most-installed skill registry in the entire ecosystem. The anthropics/skills repo spans design, document generation, and developer tooling. See Official Skills from Technology Makers for the full breakdown.

Terminal window
npx skills add anthropics/skills

Key backend skills in this registry:

  • mcp-builder (29K installs) — A comprehensive guide for creating high-quality MCP servers. If you are building custom tooling that Claude Code or Cursor will interact with, this skill teaches the patterns for reliable server implementations.
  • webapp-testing (36K installs) — Playwright-based testing patterns for local web applications. Teaches the AI to write end-to-end tests that cover real user flows rather than shallow unit tests that miss integration issues.
  • claude-api (8.8K installs) — Claude API integration patterns for building AI features into your backend.

Anthropic Claude Code Skills (9.4K installs)

Section titled “Anthropic Claude Code Skills (9.4K installs)”

The official Claude Code skill registry covers the tool’s own extension points.

Terminal window
npx skills add anthropics/claude-code

Key skills in this registry:

  • mcp integration (827 installs) — Teaches the AI how to integrate MCP servers via SSE, stdio, HTTP, and WebSocket transports. Essential when your backend needs to connect to external MCP services.
  • hook development (745 installs) — Event-driven hooks for Claude Code. Teaches patterns for pre-commit validation, post-generation linting, and custom automation triggers.

The official Cloudflare skill registry is packed with skills for edge-first backend development. If you deploy to Workers, this is mandatory.

Terminal window
npx skills add cloudflare/skills

Skills in this registry:

  • wrangler (976 installs) — The Cloudflare Workers CLI and all platform primitives: KV, R2, D1, Vectorize, Workers AI, and Containers. Teaches the AI to write correct wrangler.toml configurations and use bindings properly.
  • cloudflare (573 installs) — DNS management, analytics, Workers monitoring, and security configuration. Covers the operational side of running on Cloudflare.
  • durable-objects (520 installs) — Stateful coordination on Workers: RPC patterns, embedded SQLite, alarms, and WebSocket handling. Critical for anything that needs server-side state at the edge.
  • agents-sdk (404 installs) — Build AI agents that run on Workers. Covers agent lifecycle, tool use, and orchestration patterns.
  • building-mcp-server-on-cloudflare (317 installs) — Build remote MCP servers with OAuth authentication deployed to Cloudflare Workers. The bridge between MCP tooling and edge infrastructure.
  • sandbox-sdk (64 installs) — Sandboxed code execution on Workers for safely running untrusted code.

Supabase Postgres Best Practices (13.9K installs)

Section titled “Supabase Postgres Best Practices (13.9K installs)”

The second most-installed backend skill in the entire ecosystem. Teaches the AI how to write performant Postgres queries, design schemas, and avoid the performance pitfalls that plague production databases.

Terminal window
npx skills add supabase/agent-skills

What it teaches (via supabase-postgres-best-practices, 13.3K installs):

  • Schema design patterns for Postgres
  • Index strategy and query optimization
  • Row-level security (RLS) policies
  • Connection management and pooling
  • Migration patterns and rollback strategies
  • Common performance antipatterns and how to avoid them

Better Auth Best Practices (12.9K installs)

Section titled “Better Auth Best Practices (12.9K installs)”

One of the most popular backend skills. Teaches the AI how to implement authentication properly using the Better Auth TypeScript framework.

Terminal window
npx skills add better-auth/skills

What it teaches (via better-auth-best-practices, 8.8K installs):

  • Session management and token handling
  • OAuth provider integration patterns
  • Multi-factor authentication implementation
  • Role-based access control (RBAC)
  • Security defaults: CSRF protection, secure cookies, rate limiting on auth endpoints
  • Database adapter patterns for session storage

The official Stripe skill registry teaches the AI payment integration patterns that handle real money safely.

Terminal window
npx skills add stripe/ai

Key skills in this registry:

  • stripe-best-practices — Covers payment intents, subscriptions, webhooks, Stripe Connect, and checkout flows. Teaches the AI to handle idempotency, webhook verification, and the edge cases that cause payment bugs.
  • upgrade-stripe — Guides Stripe API version upgrades. When you need to move between API versions, this skill teaches the AI to identify breaking changes and update your integration safely.

A type-safe, high-performance backend framework skill for Bun-based projects.

Terminal window
npx skills add elysiajs/skills

What it teaches (via elysiajs, 911 installs):

  • End-to-end type safety from route definitions to response types
  • Plugin composition patterns
  • Request validation with built-in schema validation
  • WebSocket handling and lifecycle hooks
  • Performance patterns specific to the Bun runtime

For monorepo backend architectures, the official Vercel Turborepo skill teaches build system patterns.

Terminal window
npx skills add vercel/turborepo

What it teaches (via turborepo, 3.5K installs):

  • Task pipeline configuration and dependency graphs
  • Remote caching for faster CI builds
  • Workspace package organization
  • Incremental build patterns
  • Shared configuration across backend services in a monorepo

The most comprehensive security skill registry, from one of the most respected security firms. Essential for any backend that handles user data.

Terminal window
npx skills add trailofbits/skills

Skills in this registry:

  • semgrep (493 installs) — Static analysis scans and custom rule creation. Teaches the AI to write Semgrep rules specific to your codebase and run scans that catch vulnerabilities before they reach production.
  • differential-review (476 installs) — Security-focused code change analysis. Teaches the AI to review diffs for security implications: new attack surfaces, privilege escalations, and data exposure risks.
  • codeql (463 installs) — CodeQL static analysis for finding vulnerabilities. Teaches the AI to write CodeQL queries and interpret results for languages like JavaScript, Python, Go, and Java.
  • property-based-testing (456 installs) — Property-based testing with Hypothesis. Teaches the AI to write tests that generate thousands of random inputs to find edge cases your unit tests miss.
  • modern-python (254 installs) — Modern Python tooling: uv for package management, ruff for linting, ty for type checking. Teaches the AI the current best-practice Python development workflow.

Additional security-focused skills from OpenAI’s registry.

Terminal window
npx skills add openai/skills

Key backend security skills:

  • security-best-practices — Language-specific and framework-specific security review patterns. Teaches the AI to audit code against known vulnerability classes for your particular stack.
  • security-threat-model — Trust boundary analysis, asset identification, and threat enumeration. Teaches the AI to think about security architecturally before diving into code.
  • playwright — End-to-end testing patterns that complement security testing.
  • cloudflare-deploy — Deployment patterns for Cloudflare Workers and Pages with security considerations.

The Wondel.ai Skills Library includes 10 engineering skills grounded in classic software engineering books. These teach the AI established code quality principles and architecture patterns rather than framework-specific conventions.

Install all engineering skills:

Terminal window
npx skills add wondelai/skills

Or install individually:

Terminal window
npx skills add wondelai/skills/clean-architecture
npx skills add wondelai/skills/domain-driven-design

clean-code (Robert C. Martin) — Readable, maintainable code through disciplined naming, small functions, SRP, and clean error handling. Teaches the AI to write code that communicates intent, not just correctness.

Terminal window
npx skills add wondelai/skills/clean-code

pragmatic-programmer (Hunt & Thomas) — Meta-principles of software craftsmanship: DRY, orthogonality, tracer bullets, and design by contract. Teaches the AI to make pragmatic trade-offs rather than dogmatic choices.

Terminal window
npx skills add wondelai/skills/pragmatic-programmer

refactoring-patterns (Martin Fowler) — Named refactoring transformations applied systematically. The AI applies specific, named refactorings (Extract Method, Replace Conditional with Polymorphism) rather than ad-hoc restructuring.

Terminal window
npx skills add wondelai/skills/refactoring-patterns

software-design-philosophy (John Ousterhout) — Managing complexity through deep modules, information hiding, and strategic programming. Teaches the AI to distinguish deep modules (simple interface, complex implementation) from shallow ones.

Terminal window
npx skills add wondelai/skills/software-design-philosophy

clean-architecture (Robert C. Martin) — The Dependency Rule: source code dependencies point inward from frameworks to use cases to entities. Ports and adapters, hexagonal architecture patterns.

Terminal window
npx skills add wondelai/skills/clean-architecture

domain-driven-design (Eric Evans) — Bounded contexts, aggregates, ubiquitous language, and domain events. Teaches the AI to model software around business domains with anti-corruption layers between contexts.

Terminal window
npx skills add wondelai/skills/domain-driven-design

ddia-systems (Martin Kleppmann) — Storage engines, replication, partitioning, transactions, and consistency models. Teaches the AI to make informed database and data system choices based on trade-offs, not defaults.

Terminal window
npx skills add wondelai/skills/ddia-systems

system-design (Alex Xu) — Scalable distributed systems: load balancing, caching, database scaling, message queues, and rate limiting. Structured approaches for designing systems that handle real traffic.

Terminal window
npx skills add wondelai/skills/system-design

release-it (Michael Nygard) — Production stability patterns: circuit breakers, bulkheads, timeouts, retry logic, and failure cascades. Teaches the AI to build systems that fail gracefully instead of catastrophically.

Terminal window
npx skills add wondelai/skills/release-it

high-perf-browser (Ilya Grigorik) — Browser networking internals: HTTP/2+, resource hints, render blocking, connection management, and Core Web Vitals optimization from a networking perspective.

Terminal window
npx skills add wondelai/skills/high-perf-browser

Many technology companies publish official backend skills. These are covered in detail in Official Skills from Technology Makers:

SkillInstallsWhat It Does
firebase/agent-skills61,000+12 skills for Firebase: Auth, Firestore, Hosting, Genkit, Data Connect
clerk/skills26,000+10 skills for Clerk auth: Next.js patterns, webhooks, orgs, custom UI
apollographql/skills13,000+GraphQL schema design, Apollo Client/Server, federation, and Rust best practices
coinbase/agentic-wallet-skills12,300+Blockchain wallets, USDC payments, on-chain queries
clickhouse/agent-skills1,700+ClickHouse analytical database patterns
neondatabase/agent-skills20Neon serverless Postgres
planetscale/database-skills8PlanetScale MySQL branching
redis/agent-skills3Redis data patterns

Backend skills define how the AI writes code. MCP servers let it interact with the systems that code runs on:

Skill (How to Write)MCP Server (How to Do)
Supabase Postgres skillPostgres MCP or Supabase MCP
Cloudflare Workers skillsCloudflare MCP
Trail of Bits security skillsGitHub MCP (for PR security reviews)
Better Auth skillFilesystem MCP (for reading auth configs)
Stripe skillsStripe MCP (for testing webhook payloads)

AI mixes patterns from multiple backend skills. If you have Cloudflare Workers skills and a generic Node.js approach in mind, the AI may generate code using Node.js APIs that do not exist in the Workers runtime. Be explicit about your target runtime in your prompts.

Skill patterns do not match your framework version. Backend frameworks evolve. Run npx skills update regularly, and check that skill instructions match your package.json or requirements.txt versions.

Too many backend skills overload context. Cloudflare skills (6 skills), Trail of Bits (5 skills), and Anthropic skills together can fill a significant portion of the context window. Only install registries relevant to your current stack. You can always add more later with npx skills add.

Security skills flag too many issues. The Trail of Bits skills are thorough. If you are getting overwhelmed with findings, ask the AI to focus on critical and high severity issues first, then work through medium and low findings in subsequent passes.