I love CC and Codex CLI. But for Cursor the harness is the same (a bit suboptimal compared to the original, at least for Codex). So missing file/dir drag & drop etc.
Maybe it’s a good idea CLIs triggering each other for reviews etc, using skills or so.
Would like to know who uses it, and for what / why
I’ve always used the IDE chat, but for a while now I’ve been using only the CLI. Running five agents in the IDE TERMINAL is effortless, but running five agents in the IDE CHAT is impossible. After 30 minutes, everything freezes. I still use the IDE to check edited files and Git control, but I do everything through the terminal.
I also run some side tasks for several agents through Windows PowerShell.
I miss checking package usage through the CLI. I still have to open the chat window in the IDE and check the current usage.
I’d love to use aliases when running agents through the CLI to make sessions easier to find, e.g., agent --a agent_alias, because when I use “agent ls,” the session names are automatically generated from the first sentence of the session start.
And above all, deleting text entered in the terminal is a nightmare… no matter what terminal i use.
The CLI is great!! I use it equally to the desktop app. I’d like it if it had better support for images and files, in particular being able to handle image responses from the mcps would be key. Also a way to view the full output from previous commands would be great.
I want to build an app for IOS and Android more easily with Cursor. I want to build IO games entirely in cursor. I’m a super power user with cursor, and can build almost anything.
I’ve used it for blender, unity, web apps, shopify apps, three.js games, and the latest, three.js and blender web portfolio.
Really enjoying the CLI! I prefer to use the CLI exclusively as it fits neatly in my setup and workflow - would find debug mode very valuable. I Run the CLI in tmux in Ghostty - mostly a smooth experience.