Explore distributed LLM inference codebase (HyperCluster)
Takeaway: I'd start with the same broad 'explore everything and list features' prompt, since this was a pure read-only investigation and it worked fine end to end.
Stack · 1 session
safetensors sessions with coding agents
1 session on safetensors with a coding agent: the agent failed at least once in 1, and 0 finished without a human stepping in.