Get Kg Graph
Return active nodes + edges for the KG visualizer.
shop_id comes from the session via kg_read_shop_id (agents.md rule #7 — never from a request argument). Sibling to GET /kg/debug/node/{id}; it does NOT touch any existing read path.
Two scopes, same response shape:
- deliverable_id given: kg_active_edges.deliverable_id is the scope. Nodes in scope = the set of nodes referenced (from_node/ to_node) by those active edges, fetched from kg_active_nodes (which excludes retired nodes) by id-set. An unknown deliverable returns empty lists.
- deliverable_id omitted (kg-debugger-view task, 2026-07-02): the WHOLE shop's active KG — mirrors /internal/kg/graph's shop-wide query (list_all_active / list_active_for_shop, same node/edge caps), behind user-auth instead of service-auth. Requested so jeevy's view can match khimaira's shop-wide monitor (Joseph, live comparison).
Read-only. Nodes carry only observation_count (a count, not a fact value), so no financial-fact payload is exposed at this endpoint — node observations are fetched separately via the single-node endpoint, which already gates them.
CACHING (kg-debugger-view task, 2026-07-02): served from an in-process per-shop snapshot cache (TTL _GRAPH_CACHE_TTL_SECONDS) unless reload is set. meta.cached/meta.age_seconds tell the caller whether this is a fresh query or a cached snapshot, and how stale it is.
Request
- Base URL: https://api.jeevyfabrication.com (the document declares no server; this is the origin it was published from)
- URL: https://api.jeevyfabrication.com/api/v1/kg/debug/graph
- Auth: none declared
Query parameters
Deliverable UUID — scope to one deliverable's subgraph. Omit to load the WHOLE shop's active KG (may be large).
Bypass the server-side snapshot cache and recompute.
Response
Successful Response
Changes
No recorded changes to this operation across all 1 revision of this API.