Store
Shared model-store health, registry inspection, download workflows, deletion, and optimization.
📄️ Get model-store health
Check whether the configured shared model store is enabled and reachable.
📄️ Get model-store registry
Return canonical store artifacts with their complete installed-card records, verification and companion ownership, current registry and advisory status, telemetry-derived fleet cache locations and coverage, and reconciliation state.
📄️ List active store downloads
List in-progress downloads being managed by the shared model store.
📄️ Request a store download
Ask the shared model store to download and register a base or companion artifact. Optional repository, revision, file, immutable card, ownership, and artifact-role fields bind the exact requested generation; omitted base fields inherit from the current model card.
📄️ Cancel a store download
Cancel one pending or active shared-store model download. Partial files remain available for a later resumable request.
📄️ Get store download status
Return current status for a shared-store download request for one model.
📄️ Delete a model from the store
Delete a model and its shared-store artifacts from the configured model store.
📄️ Per-node storage breakdown (local node)
Return the local node's artifact inventory across staging, direct download, and configured read-only roots. Every entry includes installed identity, verification and manifest state, companion ownership, location kind, size, last use, and live-runner use, plus event-log and filesystem capacity. Reconciliation queries each node directly; operator views use fabric telemetry.
📄️ Get model-store reconciliation status
Return fleet inventory progress, pending cache imports, and the most recent reconciliation failures.
📄️ Retry model-store reconciliation
Run an immediate node-local operator rescan. The endpoint is loopback-only; automatic periodic reconciliation remains enabled.
📄️ Start model optimization
Start an optimization job for a model already present in the shared store. Use this for workflows such as OptiQ conversion or alternate artifact generation.
📄️ Get model optimization status
Return the current status of an optimization job for one shared-store model.