Persist a checkpoint snapshot.
If a checkpoint with the same id already exists it is overwritten.
The snapshot to persist.
Load a checkpoint by its unique checkpoint identifier.
The exact checkpoint id assigned at save-time.
The matching checkpoint, or null when none exists.
Load a checkpoint for the given runId.
When nodeId is supplied, returns the most recent checkpoint for that specific
node within the run. When nodeId is omitted, returns the most recent checkpoint
for the run regardless of node (equivalent to latest(runId)).
The graph run identifier.
Optional nodeId: stringOptional node filter.
The matching checkpoint, or null when none exists.
Return the most recently persisted checkpoint for a run, or null when the run
has no checkpoints.
The graph run identifier.
List lightweight metadata descriptors for all checkpoints belonging to a graph.
The compiled graph identifier.
Optional options: { Optional limit?: numberMaximum number of entries to return (most-recent-first).
Optional runOptional filter to a single run within the graph.
Array of CheckpointMetadata, sorted by timestamp descending.
Create a new run branching from an existing checkpoint.
The operation deep-clones the source checkpoint, assigns a fresh runId and
checkpoint id, applies any patchState overrides, persists the new checkpoint,
and returns the new runId.
The checkpoint to fork from.
Optional patchState: Partial<GraphState<unknown, unknown, unknown>>Optional partial GraphState overrides applied after cloning.
The new runId for the forked run.
When checkpointId does not exist.
Persistence contract for checkpoint snapshots.
Implementations may back this with in-memory maps (for testing / ephemeral runs), SQLite / Postgres rows, object storage blobs, or any other durable medium.
All methods are async to accommodate I/O-bound backends without interface changes.