Get a merge queue's dependency graph
Returns the queue’s dependency graph: which pull requests the engine considers related, and why.
This does not tell you what runs next. The engine schedules on a priority that cascades backwards through the graph and on a pending-failure gate, and neither is in this payload — priorityValue here is each pull request’s own value, not the cascaded one. For dispatch order use GET /v2/merge-queues/{id}/order.
Edges are transitively reduced, so an indirect dependency appears through the node between its endpoints rather than as its own edge. They mean overlapping impacted targets, not a stacked-branch relationship.
Only pull requests the engine has admitted to its graph appear: anything past the graph’s soft cap, or not yet ready, is absent. includeImpactedTargets adds the target lists that explain each node and edge; it is off by default because populating it costs a lookup per distinct commit.
Authorizations
An API key, sent as Authorization: Bearer <key>. An organization admin creates one in the Trunk app under Settings → Developer → API Keys and chooses the merge-queue permissions it holds; a key with none can read merge queues but change nothing. Those permissions scope merge queue only: every key can use the Flaky Tests and Dynamic CI endpoints, including their writes, whatever permissions it holds.
Path Parameters
The merge queue's id.
"xK9mP2nQ"
Query Parameters
Include the impacted-target lists on nodes and edges. Off by default: resolving them costs one lookup per distinct commit in the graph.
false
Response
The queue's dependency graph
The queue's dependency graph. The dispatch order is NOT derivable from this — the engine schedules on a cascaded priority and a pending-failure gate, neither of which is in this payload; use GET /v2/merge-queues/{id}/order. It also contains only pull requests the engine has admitted to its graph, so anything awaiting admission, or not yet ready, is absent.