schema string | generated_utc timestamp[s] | counts dict | nodes list | edges list |
|---|---|---|---|---|
szl.estate-graph/v1 | 2026-11-09T01:46:56 | {
"nodes": 369,
"edges": 97,
"by_kind": {
"repo": 134,
"model": 85,
"pr": 58,
"space": 33,
"dataset": 51,
"gpu_container": 8
}
} | [
{
"id": "gh:platform",
"kind": "repo",
"pushed": "2026-11-07T00:00:00",
"archived": false,
"stars": 0,
"lang": "TypeScript",
"title": null,
"draft": null,
"likes": null,
"status": null
},
{
"id": "gh:.github",
"kind": "repo",
"pushed": "2026-11-08T00:00:00",
... | [
{
"src": "gh:platform",
"dst": "pr:platform#1038",
"rel": "has_open_pr"
},
{
"src": "gh:platform",
"dst": "pr:platform#1037",
"rel": "has_open_pr"
},
{
"src": "gh:platform",
"dst": "pr:platform#1035",
"rel": "has_open_pr"
},
{
"src": "gh:platform",
"dst": "pr:... |
Estate Graph v1 — the living anatomy as data
369 nodes / 97 edges: 134 repos, 85 models, 51 datasets, 33 Spaces, 58 sampled open PRs, 8 GPU containers. Edges: has_open_pr, publishes (repo->HF artifact, name-matched), trains_publishes (container->model). Schema: szl.estate-graph/v1.
Regenerated on a schedule, this becomes the living anatomy: the estate as a queryable, learnable graph. Next layers: PyG HeteroData for GNN anomaly detection (incident labels from szl-org-health), and receipt edges.
- Downloads last month
- 36