Skip to content

Knowledge graph

Intelligence → Knowledge Graph (/gravity/knowledge) shows your organization’s resources and people as a set of nodes and the relationships between them, rather than as separate lists on separate screens.

The graph currently connects two kinds of things:

  • Resource nodes - your cloud instances and DNS zones, each carrying properties like provider, region, and status
  • User and organization nodes - your organization’s members, each linked to their organization by a “member of” relationship

That member-to-organization link is, today, the one relationship type the graph actually populates. The node list also supports service, policy, network, and database as node types and several other relationship kinds (depends-on, connects-to, exposes, and so on) as filter and display options, but nothing currently generates them - resource nodes appear in the graph without edges connecting them to each other or to your organization. Treat the graph today as an inventory of resources and a membership map, not a live dependency map between your services.

The graph is populated by a seed action that reads your organization’s current cloud instances, DNS zones, and members and creates or updates the corresponding nodes. It’s safe to run more than once - re-seeding updates existing nodes rather than duplicating them. There’s no automatic, continuously-updating sync behind this today: the graph reflects whatever was true the last time it was seeded, not real-time state.

TypeColor in the UIPopulated by seeding today?
ResourceBlue (info)Yes - cloud instances and DNS zones
UserGreen (success)Yes - organization members
ServicePrimaryNo
PolicyAmber (warning)No
NetworkSecondaryNo
DatabaseRed (error)No

The four types marked “no” are part of the graph’s data model and appear as filter options, but nothing in the platform currently creates nodes of those types - don’t expect to find them.

The main view is a filterable table - filter by node type using the chips above it, each showing a live count. Click any row to open a detail panel showing:

  • The node’s label, type, and resource type/ID
  • When it was last seen and whether it’s marked active
  • Any additional properties recorded for it
  • Connected To - everything one hop away from this node, found by tracing its actual relationships

Given the current state described above, the graph is most useful for:

  • “Which members belong to which organization, and in what role?”
  • “What cloud instances and DNS zones does my organization have, and what state are they in?”

It is not currently a source for “what depends on what” or “what would break if I changed this” - for actual infrastructure relationships, use Instances, Networks, and DNS directly.