Good points and indeed thinking about this quite a bit. Currently leaning towards a CLI first approach so that Claude/Cursor/[insert coding agent] can configure and control the ide. Feels a bit meta, but also makes it extremely user-friendly.
most surprising was that something this lightweight made such a big impact on my productivity. its really nice to have persistent Claude teams on my remote machines that I can always access no matter what.
I love cmux! ironically you can use tmux-ide within cmux. The idea is to make it an agent development environment that's great when ran on a remote machine :).
I've been building with AI coding tools for about two years and kept hitting the same problem: every session starts fresh. You explain your project, build something, close the laptop, then do it all again tomorrow.
Tried fixing it with better prompts. Didn't help. The issue was organizational, not instructional.
So I made three folders: context (reference files), plans (numbered prompts for multi-session work), and messages (agent coordination). No plugins, no infra, just markdown files.
It's simple but it changed how much I actually ship. Open sourced it in case others have the same problem.
It’s user facing, so will just have an option for users to regenerate the explanation. It happens rarely enough that it’s not a huge issue, and normally doesn’t effect content (I think once I saw it go a little wonky and end the sentence with a few random words). Just sometimes switches to mono space font in the middle of a paragraph, or it “spells” a word wrong (spell is in quotes because it will spell `chien` as `chi§en`).
It’s pretty rare though. Really solid model, just a few quirks
