Hi HN! I'm the creator of Ebbforge. I spent the last few months obsessed with one question: why are 'autonomous agents' so slow and expensive?

Most frameworks today are just sequential LLM wrappers. I wanted to see if I could build a pure Rust swarm engine that runs 10 million agents locally with zero API cost. I'm especially looking for feedback on our grid-partitioned parallel processing and how we're handling TD-RL divergence without drifting into chaos. AMA!