No sandboxed execution
DebugAI never runs your code. It reasons about it, so there's no language runtime to install and nothing untrusted gets executed on either end.
about · v1.0
A debugger for the gap between what you meant and what you wrote — built by one developer who was tired of debuggers that only tell you what the computer already knows.
Most tools that claim to help you debug only look at the code. A linter catches typos. A stack trace tells you where something threw. Neither one knows what you were trying to build in the first place — so neither one can tell you when the code runs fine and still does the wrong thing.
DebugAI starts from intent. You describe what the code is supposed to do, paste the code itself, and add the console output if there is one. It finds the mismatch between those three things, explains it in plain language, and hands back a fix you can paste straight in.
DebugAI never runs your code. It reasons about it, so there's no language runtime to install and nothing untrusted gets executed on either end.
The API is a small Workers function — no server to manage, cold starts measured in milliseconds, and requests handled close to wherever you are.
Plain HTML, CSS and JavaScript — no framework, no build step. Every file does one job, which keeps the whole thing easy to read and easy to change.
Paste and go. DebugAI reasons about code the same way a person reading it would, so it isn't tied to one language or ecosystem.
DebugAI is built and maintained by Nirmal, a self-directed frontend developer who's also learning backend development from the ground up. It's one of several self-contained web projects — see the rest on the portfolio site.
View source on GitHub