Prompt Like a Pro

3 simple tips to improve your prompts

Claude 4 is currently my favorite AI model for coding and I recently attended an Anthropic workshop (the creators of Claude) and picked up some genuinely useful tips on writing better prompts. If you’re using LLMs like Claude for research, debugging, or architectural decisions, a few simple tweaks can make a big difference.

Here’s what I learned:

---

🔍 1. Let it think out loud

When tackling complex problems, like understanding a bug, analyzing code, or designing a system, you’ll get better answers if you ask Claude to explain its reasoning.

Try adding “Think step-by-step.” to the prompt.

This triggers a chain of thought, helping Claude walk through the problem logically.

👉 If it’s not writing its thought process, it’s not actually thinking!

---

🎯 2. Give it a role

Framing matters. Instead of just asking for help, assign Claude a specific perspective.

Example:

“You are a senior backend engineer reviewing this TypeScript function for edge cases.”

This gives it context to tailor the response more effectively.

---

📚 3. Use examples

LLMs learn from patterns. If you want Claude to follow a specific style or structure, include a few short examples.

Example:

Provide two properly formatted React components, then ask:

“Write a new component following the same structure and naming conventions.”

---

These small changes can turn vague or generic outputs into something much more useful.

Hope these tips help you get more out of your AI assistant, whether you’re debugging, designing, or just stuck on a tricky problem.

Until next time, happy coding ✨

—Elsa

P.S. Check out the official Anthropic website for more tips.