Resources and Relationships

How Knowledge Tree represents infrastructure as resource nodes and typed relationship edges in the knowledge graph.

Resources

Every discovered object is a resource. Resources are stored as labeled nodes in the knowledge graph with a set of properties. Each resource has a globally unique ID, a type in dotted notation, and key-value properties.

Resource Fields

Field	Description	Example
`id`	Globally unique identifier	`aws:ec2:us-east-1:123:i-abc123`
`type`	Dotted resource type	`aws.ec2.instance`
`name`	Human-readable name	`prod-api-server`
`provider`	Cloud provider	`aws`, `azure`, `gcp`, `kubernetes`
`region`	Geographic region	`us-east-1`, `europe-west1`
`account_id`	Account/project ID	`123456789012`
`properties`	Key-value attributes	state: running, instance_type: m5.large
`tags`	User-defined labels	env:prod, team:platform
`raw_data`	Full provider JSON	Preserved for detailed inspection

Resource Types

Resource types follow a provider.service.resource naming convention. Knowledge Tree recognizes types across four providers:

AWS

aws.ec2.instance        aws.rds.instance       aws.s3.bucket
aws.lambda.function     aws.eks.cluster        aws.elb.load_balancer
aws.vpc                 aws.subnet             aws.nat_gateway
aws.sqs.queue           aws.sns.topic          aws.dynamodb.table
aws.elasticache.cluster aws.iam.role           aws.iam.policy
aws.cloudfront.distribution aws.apigateway.rest_api

Azure

azure.compute.virtual_machine   azure.sql.database
azure.web.app_service           azure.web.function_app
azure.containerservice.managed_cluster
azure.storage.account           azure.network.virtual_network
azure.network.load_balancer     azure.keyvault.vault

GCP

gcp.compute.instance     gcp.cloudsql.instance   gcp.storage.bucket
gcp.container.cluster    gcp.run.service         gcp.compute.network
gcp.pubsub.topic         gcp.bigquery.dataset    gcp.spanner.instance

Kubernetes

k8s.namespace           k8s.deployment          k8s.service
k8s.pod                 k8s.ingress             k8s.config_map
k8s.secret              k8s.persistent_volume_claim
k8s.daemon_set          k8s.stateful_set        k8s.cron_job

Custom Resource Types

Custom discovery plugins can emit any resource type following the dotted notation. Unknown types are stored with provider set to "unknown" and are still fully queryable.

Relationships

Relationships connect resources as typed, directed edges in the graph. They represent how resources depend on, connect to, or contain each other.

Relationship Types

Type	Description	Example
`DEPENDS_ON`	Runtime dependency	Service depends on a database
`CONNECTS_TO`	Network connection	App connects to a cache endpoint
`ROUTES_TO`	Traffic routing	Ingress routes to a service
`RUNS_ON`	Execution hosting	Pod runs on a node
`EXPOSES`	Public endpoint	Load balancer exposes a service
`OWNS`	Ownership/containment	VPC owns a subnet
`CONTAINS`	Physical containment	Cluster contains a node pool

Properties

Resource properties are stored as key-value pairs prefixed withprop_ in the graph. For example, an EC2 instance would have prop_state, prop_instance_type,prop_private_ip, etc. These are fully searchable via Cypher queries.