@poteto's team at @aiDotEngineer: 11,000 PRs to production in October, ~227 contributors, revert rate around 0.5%, a stable client release roughly every day. She also said getting to real trust in agent output took her about six months. Those two numbers belong together.
@simonw's year in LLMs at @aiDotEngineer: coding agents became a daily tool after Claude Opus 4.5 and GPT-5.1 landed in November 2025. A year later he had Opus 5.5 run for 12 hours, playtesting about 1,000 simulated battles to make a game more fun.
Cursor's breach didn't start in their code. It started with a vendor's leaked key to a shared bucket of coding agent traces, full of plaintext credentials.
Agent traces are the new source code. Redact secrets, lock down storage, audit every vendor with access.
We’ve published an update on our recent security incident. See a detailed overview of what happened and our remediation steps here: cursor.com/blog/security-…
The line I'd keep from @dexhorthy's anti-slop manifesto: own the intent, taste and consequences of what you ship, and decide how much to review based on what's at stake. slopmanifesto.com@aiDotEngineer