Add the root --gpu broker, strict GPU entry points, daemon connect-or-start orchestration, process-group isolation, bounded retry, named child reaping, and a deterministic headless lifecycle probe. Cover the complete launch matrix with real subprocess acceptance, make root Cargo runs unambiguous, and document the coherent build and one-command workflow.
Record the approved managed-launch contract, daemon lifecycle and process-group rules, strict GPU CLI behavior, headless acceptance seam, and the complete acceptance matrix before implementation.
Levi Neuwirth