How do you manage multiple agents?

I know many of you use agent decoding tools such as Claude Code or Cursor, and I wanted to know how you manage using multiple agents together, not just sub-agents, in your workflows. I originally started off using multiple agents in Claude Code, which meant I had to use the sidebar and switch between many agents. This got very confusing because I would forget which one is stuck, which one is doing.

I shifted into a warp terminal, where I then split my screen and assigned an agent to each quadrant of the screen. However, I still found it hard to memorise, and it was quite sore and confusing on the eyes. I was wondering what you would think of a canvas being used to manage all your agents, a canvas where agents are linked together. You can close and open an agent if you're sure that you don't need it, or you know what it's doing, with obvious naming, so you know what it's doing. There should also be clear indicators when an agent is stuck or needs you to intervene.

I'm building a product to help with that. It's called Skribbl. I just wanted any feedback on the site.

For memories I use YantrikDB MCP server which connects all my workspaces among distributes systems.

To collaborate runtime with workspaces I use https://github.com/spranab/swarmcode