## Problem Need 6 H100 GPUs initially (2 staging, 4 production), to get the scope remote inference Infra established. ## Proposed Solution <!-- How should it work? Include mockups or diagrams if helpful. --> ## Alternatives Considered <!-- What other approaches did you think about? --> ## Scope - [ ] Shell / core platform - [ ] Plugin SDK - [ ] Existing plugin: <!-- name --> - [ ] New plugin - [ ] Infrastructure / CI - [ ] Documentation