/api/v1/graph. It has three read routes and no writes.
How it works
- You call a graph route with your API key.
- Nexio assembles the whole graph for your organization from its own records at request time.
- The route filters the assembled graph (all of it, one node kind, or one node and its links) and returns it with
generatedAt, the time the graph was assembled.
Nodes and links
A node is one thing in your data estate, and its kind says what sort of thing. Node kinds are a closed registry of 16, fixed in code:connection, warehouse_census, schema_group, warehouse_table, corpus, doc_type_group, document, landing_plane, landed_table, canon, served_table, reference_store, live_source, derivation, engine and api_surface. The API serves 14 of them; document and warehouse_table answer 400 unsupported_node_type. Of the 14, the graph emits two kinds today.
The other 12 served kinds return an empty
nodes array today.
A node key is <kind>:<id segments>, for example connection:5f0c9a52-1d2b-4c47-9a0e-3b8d6f1e2a47. The routes take the kind and the rest of the key as two path segments.
A link (an edge) joins two nodes. Today every edge is a feeds edge from a connection to a derivation scheduled on it. evidence.statement says in one sentence why the link exists: an active schedule, or a paused one. observed is false, because the link comes from configuration rather than recorded activity.
The whole-graph response also carries findings. It is an empty array today.
What you call
Credential: the organization’s live API key, or a scoped key with
graph:read. A sandbox or test organization key gets 403 scoped_key_required. See Authentication and access.
Example: list your connections
200 OK
Example: one node and its links
200 OK
Errors
Next
Connections
What a connection is and how Nexio configures one.
Get the whole graph
Endpoint reference.