This website uses cookies
Read our Privacy policy and Terms of use for more information.
AI
+1
Nov 10, 2026
•
5 min read
GitHub's new storage design says almost everything in a Git push needs no coordination at all. Here is the split, why it matters for your own write-heavy systems, and a short toy that shows the numbers.
Oct 31, 2026
AI deskilling is not a vibe. It has been measured in colonoscopy rooms, flight simulators and a randomized coding trial. Here is what those numbers actually say, and the one metric your team needs to start tracking.
Oct 28, 2026
6 min read
Anthropic told its model it was offline. It was not, and the model hit real systems. Why a prompt is not a network control, plus a probe you can run from inside your own agent sandbox.
Oct 13, 2026
OpenAI withdrew three math papers in under 24 hours, an Epoch benchmark caught two frontier agents inflating their own results, and Lean's soundness bugs keep surfacing. The pattern is the same each time, and it applies to your CI pipeline.
Jul 15, 2026
8 min read
AI Now's Friendly Fire exploit turns Claude Code and Codex into the delivery vehicle for the malware they were asked to find. The reason it works is written plainly in Anthropic's own design doc.
Jun 27, 2026
The "MCP is bloated, rip it out" crowd is right about the bill and wrong about the cause. The token tax isn't the protocol. It's a convention nobody questioned: dumping every tool definition into the model's context.
Jun 15, 2026
7 min read
Teams keep defending agents the way they learned to defend web apps in 2008. That instinct is wrong, and the difference is structural, not a matter of trying harder.
Coding
+3
May 22, 2024