Rahul

For the past few months I've been juggling three AI coding agents Claude Code for architecture, Codex for implementation, GLM for review. The pain wasn't any single one of them. It was the switching cost.

Every time I switched engines I had to re-explain the project: which framework, which conventions, where the old conversation left off. None of them remembered what the others knew. I'd waste 10 15 minutes per switch re-loading context, and half the time I'd just give up and stay on whatever I opened first.

So I started building a small thing for myself. Local desktop app. Lets me run all three engines from one window, with a shared memory layer underneath.

The thing I underestimated: persistent memory is the whole game. Once Claude and Codex can both see "this project uses SwiftData + strict concurrency + @Model on iOS 18+," switching becomes free. I stopped caring which engine I was on and started picking per-task.

not talking about the obvious stuff like "don't rm -rf prod", more the gray area tasks where the agent is probably right but you still want eyes on it before it goes out.

for me it's anything that changes a pricing table or a billing calculation. I'll let an agent refactor half my codebase overnight, but the second a PR touches how much someone gets charged, I read every line myself even if the diff is three characters. one off-by-one there and it's not a bug report, it's a support ticket and a refund.

honest self-assessment time. when an agent finishes a task and shows you a diff, what actually happens?

for me it depends entirely on the size. under ~30 lines I read every line. past that I mostly skim for anything that touches auth, payments, or migrations and trust the rest if tests pass. I know that's not a great policy, it's just the realistic one given how much code moves through these sessions in a day.