Meaning Diff Detects Test-Worsening Patches

Meaning Diff uses Jev to flag patches that make tests pass by weakening or bypassing them. In a broken checkout, it caught a hidden payment failure, a weakened assertion, and unconditional receipt-delivery success.

View on X time343ms

John@johnsandovaI𝕏

An AI agent can make the tests pass by making the tests worse. I gave one a broken checkout. It hid the payment failure, weakened the assertion, and replaced receipt delivery with unconditional success. Meaning Diff flagged all three in 343ms using Jev. Open source. Looks at the meaning of the patch, not just the lines. Link Below

Sep 22, 2026X postsView on X

The author describes Meaning Diff as open source. It evaluates the meaning of a patch, not just its lines.

Also filed under Tools & apps

Labels LinkedIn Posts for AI Slop
Slop Radar labels LinkedIn posts as human, unclear, or AI slop as users scroll, with reasons shown on hover. It is a Chrome extension powered by Jev.
Hands-Free Recipe Page Control
Recipe Mode is a Chrome extension powered by Jev that lets users cook hands-free on any recipe page using commands such as “next,” “how much butter?” or “set a timer.”
Helm Release Risk Estimator
Jelm is a Jev-powered Helm release risk estimator for Kubernetes CI/CD. It scores deployment risk, analyzes rollouts, gates Gateway API canaries, and supports safer promote/rollback decisions.
Quilt Tools and JEV Experiment Lab
SuperInstance’s quilt-tools repository contains ten working Quilt tool prototypes and a springboard lab for JEV/MothQuantum experiments.