AI Developer Cookbook

Your team just adopted AI coding tools and everyone is asking the same question: “What do I actually type into the prompt box?” This cookbook answers that with 500+ battle-tested, copy-paste prompts organized by technology stack. No theory, no fluff — just recipes that work when you paste them into Cursor, Claude Code, or Codex.

What You Will Walk Away With

• Ready-to-paste prompts for every major framework and language
• Recipes that work across all three tools with tool-specific variations where they matter
• Real scenarios pulled from production codebases, not toy examples
• Expected outputs so you know what good looks like

How Recipes Are Structured

Every recipe in this cookbook follows the same format:

1. Scenario — the real-world problem you are solving
2. Prompt — the exact text to paste into your tool, inside an Aside block so it stands out
3. Expected output — what the AI should produce, so you can verify quality
4. Tool variations — Tabs showing Cursor, Claude Code, and Codex differences where the workflow diverges

Note

Most prompts work identically across all three tools. We use Tabs only when the workflow genuinely differs — for example, Cursor’s inline edit vs Claude Code’s terminal-first approach vs Codex’s multi-surface workflow.

Recipe Categories

Frontend Recipes React, Vue, Nuxt, Angular, Next.js, and mobile UI patterns -- component generation, state management, performance optimization

Backend Recipes Node.js, Python, Elixir, PHP, Java, Go, Rust, API design, and serverless -- from CRUD to distributed systems

Mobile Development Expo, Flutter, and React Native recipes for cross-platform mobile apps

Database Recipes SQL tuning, NoSQL modeling, ORM patterns, and migration strategies with database MCP servers

DevOps Recipes Docker, Kubernetes, CI/CD pipelines, infrastructure as code, and monitoring with AI assistance

Error Handling Recipes Systematic debugging, production logging, and error recovery patterns

Quick Start — Your First Recipe

Pick your stack and paste a prompt. Here is a taste of what each section offers:

Cursor

Open Agent mode with Cmd/Ctrl + I and paste any recipe prompt. Cursor will execute file edits, run terminal commands, and iterate until the task is complete. Use Cmd/Ctrl + K for inline edits on a single file.

Claude Code

Run claude in your terminal and paste the prompt. Claude Code will read your project files, execute shell commands, and make edits across multiple files. Use /init first to generate a CLAUDE.md with project context.

Codex

Use Codex CLI (codex), the Codex app, or Codex inside your IDE. Paste the prompt in any surface — Codex uses worktrees for safe parallel execution and integrates with GitHub, Slack, and Linear for automated workflows.

Tip

Your first recipe to try right now:

Review my project structure and generate a CLAUDE.md (or .cursorrules, or codex.md) file that documents: the tech stack, key architectural patterns, testing conventions, file naming rules, and any domain-specific terminology. Read at least 10 source files before generating.

Tips for Getting the Most Out of Recipes

• Adapt, do not blindly paste. Replace technology-specific details (database names, API endpoints) with your own, but keep the prompt structure intact.
• Chain recipes together. A backend API recipe followed by a testing recipe followed by a Docker recipe gives you a complete feature pipeline.
• Start with the model that fits. Claude Opus 4.6 for complex multi-file tasks. Claude Sonnet 4.5 for fast iteration. Gemini 3 Pro when you need massive context. GPT-5.3-Codex for Codex-native workflows.
• Use MCP servers for live context. Database MCP servers, Figma MCP, GitHub MCP — these give the AI real data instead of guesses.

What Is Next

Pick a category above and start cooking. Every recipe is self-contained — you do not need to read them in order. If you are new to AI-assisted development, start with the React patterns or Node.js patterns for the most comprehensive introductions.