Back to blog
Coding7 min read

Using AI to debug a Next.js project without creating chaos

A practical way to ask Edvin for debugging help that preserves context, narrows the issue, and avoids random refactors.

Next.jsDebuggingCoding

The short version

The best AI debugging prompt gives the symptom, expected behavior, recent change, error output, and files involved. Then ask for a narrow fix, not a redesign.

Share the failure shape

Tell Edvin what you expected, what happened instead, and what changed recently. Error output is useful, but context tells the assistant where to look.

If you have a screenshot, route, or exact command, include it. Debugging gets faster when the failure is concrete.

Set scope boundaries

Say what not to change. For example: fix the streaming bug, do not rewrite the chat UI, do not touch unrelated routes, and keep Supabase persistence intact.

This is not politeness. It is engineering hygiene. AI tools are more useful when the blast radius is explicit.

Ask for verification

A fix is not done until it is checked. Ask Edvin to run typecheck, lint, build, or the specific reproduction command.

If a test fails for unrelated reasons, the assistant should say that clearly instead of quietly changing old code.

Apply this in Edvin

Turn the article into a working prompt, then save the result as a project asset.

Help me apply this Edvin.ai article to my current project: Using AI to debug a Next.js project without creating chaos

Use this prompt