Ever since the new Claude Code model launched around Christmas, I’ve been in love. In fact I’ve been using it so much that I’ve stopped using OpenAI / ChatGPT altogether.
One of the things I really wanted to try is to see if I can spawn multiple agents in Claude Code to make the tasks that I’m doing go faster.
The default is to use sub-agents (in one process) but I wanted it to spawn a team of agents where you can interact with each of the agents independently as well. The feature is actually experimental and is documented here - I wanted to actually see it rather than just believe it.
This requires a little bit of jiggery pokery with tmux and a bunch of environment variables (let me know if you want the details), but I was able to get Claude Code to do a research task in which case by launching four teammates. In this baby example, for the hottest AI companies, I look for the tweets from the last week from each of the leaders.
Once you see this there’s no going back. [I know all the developers already know this but this is more for the unwashed non-developer masses!]
