This looks promising guys , congratulations , i am leading my team to switch to CLI instead of VS code extensions as you stated we are working on different IDEs and stacks , we will give it a shot and see how it goes
Well I’ve been using it for a while now something like “screen” would be nice so even when I lose my connection the cli continues to work, screen is ok but sometimes it’s difficult to use when text is rolling up, so as a feature for kilo would be better
Building on OpenCode instead of rolling your own foundation is the right call. The AI CLI space is fragmenting fast — Claude Code, Codex CLI, Aider, OpenCode, now Kilo — and the ones that win long-term will be the ones that don't force you into a single model provider.
The real test for me as someone who lives in the terminal (Claude Code + tmux all day) is whether model-switching actually works seamlessly in practice. Using Sonnet for quick edits and switching to Opus for architectural decisions is the workflow I want but haven't found executed well yet. Curious how Kilo handles that context handoff between models mid-session.
haven't had a chance to run it in a real sprint yet but planning to this week. will report back
first impressions from the docs: the deviation log concept looks useful. when agents drift from the plan (and they always do), having that record is gold for debugging
It's the era of AI, time to use CLI & TUI.
I used Kilo CLI, and it's great!
Thank you so much!!
Glad to hear you liked it! What kind of projects are you building?
I'm building some cool apps https://taskdone.gateandtech.in/
I have built my portfolio only using Kilocode CLI: https://lakshman-portfolio-beta.vercel.app/
Just build some awesome cool things doe all the resources: https://awesome-horizon.vercel.app/
https://awesome-skills-eight.vercel.app/
Now learning to develop how to build CLI like Kilo & Opencode.
Thank you so much!
This looks promising guys , congratulations , i am leading my team to switch to CLI instead of VS code extensions as you stated we are working on different IDEs and stacks , we will give it a shot and see how it goes
Oh yes, it’s great is less cluttered and is feeling better while on the way, congrats on the new version
Well I’ve been using it for a while now something like “screen” would be nice so even when I lose my connection the cli continues to work, screen is ok but sometimes it’s difficult to use when text is rolling up, so as a feature for kilo would be better
Thanks for the feedback! Have you tried the new version? If so, how does it compare to our previous version?
Building on OpenCode instead of rolling your own foundation is the right call. The AI CLI space is fragmenting fast — Claude Code, Codex CLI, Aider, OpenCode, now Kilo — and the ones that win long-term will be the ones that don't force you into a single model provider.
The real test for me as someone who lives in the terminal (Claude Code + tmux all day) is whether model-switching actually works seamlessly in practice. Using Sonnet for quick edits and switching to Opus for architectural decisions is the workflow I want but haven't found executed well yet. Curious how Kilo handles that context handoff between models mid-session.
Thanks for the support! I'd love to hear what you think after you use it in practice.
haven't had a chance to run it in a real sprint yet but planning to this week. will report back
first impressions from the docs: the deviation log concept looks useful. when agents drift from the plan (and they always do), having that record is gold for debugging
Let us know what you think when you do!
Glad to hear that! That's the power of open-source - you can build better tools much faster when you can build alongside a massive community!