#testing

I wasn't looking for a performance problem. I wanted to see what happened when the network got slow...

I deleted !arrays || from a guard in radashi, a TypeScript utility library that gates CI on 100% line...

Last week I compared two CSV snapshots of a price table, one export from staging, one from...

Why Your D365 F&O Automation Breaks: The Hidden Logic Behind TargetId and RootId If...

In 2026, the central challenge of moving Generative AI into production isn't what the model can do,...

Update — v0.3.1 released. CauterRule is now live on GitHub and PyPI. It turns repeated agent...

Six weeks of integration work with a new settlement provider, a sandbox we could hit as often as we...

Using agent-browser for goal-oriented UI exploration, independent assertions for verification, and turning discovered failures into cheaper regression checks.

Row counts hide the two failures that hurt: rows that vanish and rows that appear. Put every row in exactly one bucket, and refuse to report if the buckets don't reconcile.

The Flash Attention 2 path was easier to finish after the algorithm existed in plain JavaScript...

TL;DR: I set out to compress a shared whiteboard for a connection that's slow (~120 kbps practical)...

Agent patches often go green by making invalid input succeed. That is not a fix. It is a silent spec...