I built FixProof to verify whether AI-generated bug fixes are actually proven

Description and use

FixProof verifies that AI‑generated bug fixes are truly proven by checking test failures before the change, passes after, and inspecting integrity warnings, yielding a reliable verdict.

Problem

AI coding tools can generate a fix and claim the bug is solved, but the regression test may not truly prove the fix.

Outcome

FixProof provides a four‑outcome verdict that ensures regression tests truly demonstrate a fix, reducing false positives from AI‑generated code changes.

Workflow

• Run the selected regression test against the committed code before the fix
• Run the same test against the committed code after the fix
• Inspect test‑integrity warnings such as removed, skipped, or weakened assertions
• Determine the verdict: verified, failed, inconclusive, or needs‑review

Evidence

FixProof follows a simple proof chain: Fail before → pass after → inspect test-integrity warnings → verdict

Source

FixProof reports one of four outcomes: verified, failed, inconclusive, needs-review

Source

FixProof v1.0 currently supports: Node.js node:test, Jest, Vitest, pytest, explicitly configured trusted project commands

Source