Before AI, I felt like I could review PRs and, based on whose code I was reviewing, I could be more trusting and casual with my review, since I knew they usually wrote good code and paid attention to details. Now, with AI, no one pays attention to details, so I have to deeply scrutinize every PR, which takes twice or 3x the effort.

AI agents almost always add things that aren’t necessary, or they follow anti-patterns based on the surrounding legacy code context.

AI coding agents get the same instructions no matter what file they're touching. So they guess, break things, then spend 30 calls fixing it.

I built sctx. It scopes context to specific files using AGENTS.yaml. The agent only sees what's relevant.

Open source, works with Claude Code today. https://sctx.dev

#Claude #LLM #AI

what I say: big plans this wknd

what I mean:

Windows 11 authentication is just bizarre. You are forced to enter three security questions for account recovery when you create a new user.

That's incredibly stupid and poses an actual security risk 🤦

Please, if you do this: don't pick actual answers to these questions. Instead enter a random token, write it down and store it safely.