FleetLM

How to produce a performance or cost number for FleetLM that is safe to publish. Use before adding any throughput, latency, cost, or efficiency claim to the README, a benchmark file, or a post.

Start a local FleetLM fleet and drive real work through it - orchestrator, one or more nodes, an interactive request, and a batch. Use when asked to run, demo, or verify FleetLM end-to-end, or after changing the orchestrator, node agent, or wire protocol.