Currently when streaming text back from the remote LLM MindMac pegs 1 CPU core at 100%.
I suspect it's a bug as I don 't think it should need this much CPU for streaming text and I suspect there is also a single thread bottleneck somewhere as it's 100% of 1 core.