Packs

A Pack is how a use case gets installed. It bundles everything specific to a job — the data model, analytics, alerts, and dashboards — into one manifest you install into your organization.

What a Pack installs

A pack manifest can seed all of these into your org at once:

• SensorProfiles and DeviceProfiles — the data model for the use case
• AnnotationClasses — the labels you'll capture
• JobDefinitions and ReferencePairs — the analytics that run on a schedule
• AlertRules — the conditions worth being told about
• MediaCaptureRules — when to record clips
• Dashboard templates — the views you start with

Installing is idempotent per organization, applied in dependency order, and reversible. This is what keeps the platform vertical-neutral: the use case lives in the pack, never in the platform code.

The bridge to use cases

Because a pack is just data, adding a new vertical means writing a new manifest — not changing the platform. The use cases section is one page per pack. Real examples ship in the repository, including a distillery angels'-share pilot and a conveyor-belt monitor.

In the API

• POST /api/v1/packs/install — install a manifest (YAML or JSON)
• GET /api/v1/packs — list installed packs for your organization
• GET /api/v1/packs/{id} — inspect one
• DELETE /api/v1/packs/{id} — uninstall