Capture session templates

Use this guide when the HEAT Capture app or protobuf uploads feed an agnostic-input node.

Canonical graph


agnostic-input
├── capture-discovery → discovery-dashboard     (debug leaf)
└── system-protobuf-extract (scoped) → … → heat-system-next-dimension

The discovery branch profiles formats, proto census, origin_uuid, and event names. Keep it a leaf: mainline analytics should not depend on discovery output.

Platform nodes

Node	Role
`capture-discovery`	Incremental profiler; emits `$heat-dataservice` for a discovery dashboard
`system-protobuf-extract`	Scoped extract: `EntityDataEvent`, `EntityState`, or `SimulationFrame`
`system-arbex-js`	Custom merge, KPIs, dataservice
`heat-system-next-dimension`	Publish Next dashboard

Enable incrementalHydration on streaming capture transforms.

Multi-user lanes

When several capture origins or simulation entities share one session, use captureIdentity session metadata and filterEntityIndex / filterOriginIndex on parallel extract branches. See Capture identity lanes.

Multi-origin sessions

One session may contain frames from multiple capture machines (origin_uuid). Keep per-record origin in extract outputs; aggregate in downstream nodes, not in a single bloated hydrate step.

Optional core-utils

When core-utils is registered, hydrate-protobuf-v2 and simulation-frame-merge can supplement platform extracts. Document the dependency; do not require it.

Shipped presets

Preset	Purpose
`heat-capture-discovery`	Discovery dashboard only
`heat-capture-mainline-scoped`	Discovery leaf + identity registry + entity lanes + mainline