I built FixProof to verify whether AI-generated bug fixes are actually proven
AI coding tools can generate a fix and tell us the bug is solved. But there is an important question: Did the regression test actually prove the fix? That question led me to build FixProof , an MIT-licensed open-source developer tool for checking the evidence behind AI-generated bug fixes. The idea FixProof follows a simple proof chain: Fail before → pass after → inspect test-integrity warnings →…
FixProof, an MIT-licensed open-source developer tool, was created to verify whether AI-generated bug fixes are actually proven. The tool follows a simple proof chain: fail before → pass after → inspect test-integrity warnings → verdict. To use FixProof, four regression tests are run: one against the code before the fix and one after.
A passing test after the change is not enough, as FixProof also checks for warning signs such as removed or skipped tests, changed assertions, or incomplete test evidence. FixProof reports one of four outcomes: verified, failed, inconclusive, or needs-review. Supported test runners include Node.js node:test, Jest, Vitest, pytest, and GitHub Actions.
FixProof does not claim to guarantee an entire bug-free application but only verifies if the selected regression evidence demonstrates the particular fix within the tested scope and environment. The tool is open source and the project is seeking developers who use AI-assisted coding tools to provide feedback on real bug fixes.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
