Structured, verified SKILL.md context files that teach AI coding agents how to work with real APIs

🔍 Searchable Directory · ⚡ Before vs. After Code · 💻 CLI Reference · 🚀 Quick Start · 📚 Skill Directory · 🤝 Contributing

---

💡 What This Is & Why It Matters

When an AI coding agent (Claude Code, Cursor, Codex CLI, Gemini CLI, etc.) writes code for modern APIs, it frequently makes critical mistakes:

• Reaching for outdated SDK methods present in old training data.
• Parsing raw request bodies with express.json() before validating Stripe webhook signatures, breaking cryptographic checks.
• Using deprecated authentication helpers (e.g. @clerk/nextjs v4 authMiddleware instead of v5 clerkMiddleware).
• Calling transactional email endpoints inside unbatched loops, hitting HTTP 429 rate limit errors.

awesome-api-skills fixes this by providing 101 curated, schema-validated SKILL.md context files. Dropping a skill file into your agent's skills directory (.claude/skills/, .cursor/skills/, .agents/skills/) gives the model exact, current API instructions, preventing guesswork before code is written.

---

⚡ The Difference A Skill Makes

Here is how an AI agent performs with and without SKILL.md context:

Scenario	Without SKILL.md (LLM Guessing)	With SKILL.md (Context Injected)
Stripe Webhook Verification	Parses JSON first (express.json()), ruining the raw buffer signature check and throwing StripeSignatureVerificationError.	Mounts express.raw({ type: 'application/json' }) on the webhook route so stripe.webhooks.constructEvent() verifies signature cryptographically.
Clerk Auth Middleware	Generates deprecated authMiddleware({ publicRoutes }) from v4, causing runtime crashes in Next.js App Router.	Uses modern clerkMiddleware() with createRouteMatcher() matching Next.js v5+ App Router specs.
Resend Batch Emailing	Iterates a for...of loop with individual send() calls, taking minutes and triggering HTTP 429 rate limits.	Packages emails into resend.batch.send([]), dispatching 500 emails in a single low-latency HTTP request.

📖 See full runnable code diffs in docs/BENCHMARKS_AND_EXAMPLES.md.

---

🚀 Quick Start

Option 1: Direct Copy into Your Project (No Installation Required)

Simply copy the skill folder for the API you are using into your agent's skill directory:

# Clone the repository lightweight
git clone --depth 1 https://github.com/ashish7802/awesome-api-skills.git

# Copy Stripe skill to Claude Code
cp -r awesome-api-skills/skills/stripe .claude/skills/

# Copy Clerk skill to Cursor
cp -r awesome-api-skills/skills/clerk .cursor/skills/

# Copy Redis skill to generic agent workspace
cp -r awesome-api-skills/skills/redis .agents/skills/

Then instruct your agent:

"Use .claude/skills/stripe/SKILL.md for payment integration."

---

Option 2: Monorepo & Local CLI Setup

For developers building tooling, validating custom skills, or managing workspaces:

git clone https://github.com/ashish7802/awesome-api-skills.git
cd awesome-api-skills
pnpm install
pnpm build

Run CLI commands locally:

# Search skills by keyword
pnpm run cli search stripe

# Validate all skill schemas
pnpm run cli validate

# Inspect workspace health
pnpm run cli doctor

---

📚 Skill Directory

Explore 101 skills across 14 core technical domains:

Category	Available Skills
Payments & Billing	stripe, paddle, lemon-squeezy, revenuecat, plaid
Auth & Identity	auth0, clerk, okta, better-auth, jwt, oauth2, openid-connect
Databases	postgresql, mysql, sqlite, mongodb-atlas, planetscale, neon, turso, drizzle, prisma
Caching & Queues	redis, redis-streams, upstash, bullmq, kafka, rabbitmq, nats
Object Storage	aws-s3, aws-dynamodb, azure-blob-storage, google-cloud-storage
AI & LLM Infra	openai, anthropic, gemini, ollama, vllm, langchain, llamaindex, pinecone, typesense, meilisearch, algolia, xquik
Backend Frameworks	express, fastapi, nestjs, hono, trpc
Frontend Frameworks	react, vue, nextjs, nuxt, sveltekit
Deployment Platforms	vercel, railway, render, fly, digitalocean, cloudflare, cloudflare-workers, deno-deploy
Infrastructure	docker, kubernetes, helm, terraform, pulumi, argo-cd, github-actions, traefik, nginx, caddy, turborepo
Observability	datadog, sentry, prometheus, grafana, loki, jaeger, opentelemetry, mixpanel, posthog
Dev Tooling	eslint, prettier, biome, vitest, playwright, git, github, xquik
Communication	slack, discord, twilio, sendgrid, resend
Platforms & CMS	shopify, mapbox, convex

🌐 Browse the interactive web catalog at docs/index.html.

---

🏗️ Monorepo Architecture

awesome-api-skills is built as a TypeScript pnpm workspace:

skills/                 101 SKILL.md files + metadata.json contracts
packages/
  cli/                  CLI binary for searching, validating, and managing skills
  core/                 Core orchestration engine and workspace management
  generator/            Pipeline generator enforcing consistent SKILL.md templates
  validator/            Zod & JSON-Schema validation engine for skills
  registry/             Graph relationship index & search resolver
  sdk/                  Programmatic Node.js SDK for accessing skill data
  shared-types/         TypeScript definitions & schemas shared across packages
scripts/
  dev/                  Validation, indexing, and static build scripts
docs/                   Searchable web directory, benchmarks, and CLI reference

---

🎯 Verification & Quality Standards

Every claim in this repository is verified by automated tests and CI checks:

1. Schema Validation: All 101 skills pass strict structural checks via packages/validator.
2. Automated Unit Tests: 31 assertions run in CI via Vitest covering all core packages.
3. Last Verified Metadata: Every skill contains a lastVerified timestamp in its metadata.json and SKILL.md header indicating when SDK signatures were verified against vendor documentation.
4. Relationship Scoring: Skill graph connections follow a transparent 5-factor hybrid scoring model (Explicit edges, Category match, Ecosystem match, Deployment match, Jaccard similarity) detailed in SPECIFICATION.md.

---

🤝 Contributing

We welcome contributions! All new skills must adhere to the standard schema:

1. Scaffold or Write Skill: Add your skill folder under skills/<skill-name>/ containing SKILL.md and metadata.json.
2. Run Validation:

   pnpm run validate:skills

3. Run Unit Tests:

   pnpm test

4. Submit PR: Open a Pull Request. GitHub Actions will automatically validate schema compliance using packages/validator.

See CONTRIBUTING.md and .github/PULL_REQUEST_TEMPLATE.md for details.

---

📄 License

MIT © 2026 Awesome API Skills Team.