Skip to main content
GET
Get the dependency graph (full org or subgraph anchored at a repo).

Authorizations

X-API-Key
string
header
required

Headers

X-Workspace-Id
string | null

Path Parameters

connected_org_id
string<uuid>
required

Query Parameters

root
string<uuid> | null

Optional repository ID to anchor a subgraph view. When set, the response is limited to the root plus its upstream and downstream neighbours within depth hops. Omit for the full org graph.

depth
integer
default:2

BFS depth from root in each direction. Ignored when root is omitted.

Required range: 1 <= x <= 10
min_confidence
number
default:0.4

Minimum edge confidence to traverse, between 0.0 and 1.0. The default 0.4 admits every reference a machine could act on — deterministic parser edges (0.9–1.0), resolver matches (0.65–0.95), heuristic references in build and CI files, package manifests and fetch commands (0.85), and references in files the classifier does not recognise (0.6–0.7), which mean 'unclassified', not 'unlikely'. It excludes only references a human reads rather than a machine executes: prose, HTML templates, generated blobs and ignore files (0.3). Pass 0.8 to restrict the answer to deterministic and build-file evidence only, or 0.0 to include everything. The value actually applied is echoed in the response, and each affected repository carries the confidence of the weakest edge on the strongest path that reached it, so a stricter answer can be recovered by filtering the response without a second request. Ignored when root is omitted: the full-org graph applies no floor at all and returns every manifest edge, in which case applied_min_confidence in the response is null.

Required range: 0 <= x <= 1

Cookies

riftmap_access
string | null

Response

Successful Response

Full dependency graph for an organization (G6 compatible).

edges
GraphEdge · object[]
required
nodes
GraphNode · object[]
required
applied_min_confidence
number | null

The confidence floor used for the subgraph traversal, or null when root was omitted. null is the honest answer for the full-org graph, which applies no floor at all and returns every manifest edge for clients to filter themselves — so a number here would misreport it.