How to use the AI Prompts for Developers
- Set your language, framework, test runner and database at the top — every prompt updates.
- Pick a prompt for your task: understand, debug, write, review, test, document or plan.
- Click Copy prompt, paste it into ChatGPT, Claude, Gemini or Copilot Chat, and replace the
[bracketed]parts with your code or details. - Ask follow-up questions in the same chat to refine the answer.
What makes a good coding prompt
- Context — language, framework and versions. “React 19 with TypeScript” gets far better answers than “JavaScript”.
- Expected vs actual — for bugs, say what should happen and what happens instead.
- The smallest relevant code — paste the function and its callers, not the whole repository.
- Constraints — no new dependencies, keep the public API, must run on Node 20…
- Output format — a diff, a table, tests only, or explanation before code.
Verify before you ship
AI assistants write plausible code that can be subtly wrong: invented APIs, missed edge cases, outdated syntax or insecure patterns. Run the code, read it as carefully as a colleague’s pull request, and ask the assistant to write tests for its own output. Never paste secrets, credentials or customer data.
Format code before pasting with the JSON formatter or SQL formatter, and check generated regular expressions in the regex tester.
Tools that pair well with the AI Prompts for Developers
- Git Cheat Sheet — Every Git command you use day to day — and the ones you need when something goes wrong — explained in one line.
- Linux Commands Cheat Sheet — The Linux and bash commands developers use on servers and laptops — files, search, permissions, processes, logs, networking and SSH.
- Docker Cheat Sheet — The Docker and Docker Compose commands you reach for every week, explained in one line.