I stumbled upon some incredibly practical advice for working with Claude Code, Cursor, and Github Copilot that I think every engineer needs to know.
Here are a few of my favorite takeaways:
➡️ Create memory files
Use files like CLAUDE.md or .cursorrules to give the AI some lasting context about your project’s structure.
➡️ Draft a solution first
Sketch out an idea yourself, even just on paper, and then use the AI to help you polish it.
➡️ Ask for different approaches
Have the LLM lay out a few different options with pros and cons so you can choose the best path forward.
➡️ Treat the AI like a partner
Explain problems to it and work together to find a solution, just like you would with a coworker.
➡️ Write detailed specs
Be super specific, even down to the exact function signatures you want the AI to implement.
➡️ Interrupt and redirect
The moment you see the AI going down the wrong path, jump in and get it back on track.
If you find these tips helpful, you should check out the full guide. It’s seriously worth the read. Here’s the link: https://coding-with-ai.dev
Feel free to repost and share this so other engineers can benefit too.