Often when an issue appears in a large file, the instinct is to throw it at an LLM immediately - especially for something like Apache ECharts.
Today I screenshotted the issue and sent it to Claude Sonnet with a ~30 word prompt.
Before Claude finished analysing the problem, I had already:
<ul>
<li>found the bug</li>
<li>fixed it</li>
<li>and reuploaded the file to the server</li>
</ul>
The fix took me about 4 minutes.
The LLM likely would have gone down a 10+ minute rabbit hole.
Reminder: not every problem should be delegated to AI. Sometimes the fastest debugger is still you.

Nigel

A developer shares a quick experience where they fixed a bug in an Apache ECharts file in 4 minutes, faster than waiting for Claude Sonnet to finish analyzing the same problem. The takeaway: reflexively delegating every debugging task to an LLM isn't always the most efficient approach — sometimes direct human debugging is faster.

Not Every Bug Needs an LLM