My agent often kicks off CI and waits for the result. It ends its turn right after the kick-off, while CI is still running, so it can respond to me immediately. Which state does that cell get: working, needs you, or done? It does not need me at that point, but it is not done either.

Hi, I'm isamu.

I often run multiple Claude Code in parallel, making it difficult to keep track of when they are idle or waiting. After trying out various terminal tools, I started to build my own.

It allows you to view multiple terminal sessions on a single screen and changes colors or status indicators based on the LLM's state. You can customize notification sounds, enabling you to perceive changes in the LLM through both sight and sound. You can launch and try it out immediately using npx. I would love for anyone running Claude Code or Codex in parallel to give it a try.

I'm co-developer, Satoshi. Here is a quick demo.

https://github.com/receptron/mulmoterminal#demo

Here is a quick instruction.

npx mulmoterminal@latest

Then, open http://localhost:34567 in your browser

[flagged]

[flagged]

undefined