Custom tools and debugging with AI

Lessons learned 4/11 on Cursor after 15 years in software

Cursor will let you improve your working environment. All those small tasks that take your time but weren't worth the effort suddenly become worthwhile.

Custom tools in minutes

You can now add a button in your admin interface for each task you used to do manually:

Verify a deployment
Retrieve logs
Count items
Download data

In a few minutes, the AI will do it wonderfully — as long as you provide a framework: which tech, which environment, which framework, which page to use.

This helps limit the gradual chaotization of your working environment while gaining fluidity.

Debugging: where it gets complicated

AI also lets you tackle this subject that terrifies so many juniors: debugging.

Simple cases

You can debug simple errors by directly copy-pasting the stack trace. The AI identifies the problem and proposes a fix in seconds.

Rare cases

For rarer errors — even mythical ones — forget it.

You need to assume that anything with few occurrences on the internet won't be natively known by the AI. For example, a specific error whose only known fix can be found in one of 4 similar error threads on GitHub has very little chance of being resolved by an agent.

In these cases, two options:

Do the research yourself and provide the appropriate resource in the prompt
Use the debugger — like any self-respecting dev

And once again, AI is a good crutch: a beginner can set it up quickly with AI, even in less-than-ideal working conditions.

Summary

Save time by creating custom tools in minutes
Bug resolution can be done automatically for simple cases
The debugger is still recommended for rare cases that AI doesn't know

Originally published on LinkedIn.