Trace ↔ Graph Alignment¶
This page explains how Semantic Execution Records (SER) align with the GraphV1 pipeline graph. The canonical mapping allows tools to reconstruct execution topology solely from SER files.
Node Identity¶
Every pipeline node is assigned a deterministic
node_uuidby the GraphV1 builder.The same
node_uuidappears: * In the canonical spec (pipeline_spec_canonical.nodes[*].node_uuid). * In each SER record asidentity.node_id.Rule: Consumers must always join SER data to the graph using
node_uuid.
SER Output¶
SER records contain minimal information. Optional fields may be supplied by the runtime and consumers should ignore unknown fields gracefully.
Pipeline & Run Identity¶
pipeline_id: deterministic hash of the canonical spec.run_id: unique per execution.Both appear in
pipeline_startrecords and in theidentityof each SER.
Trace → Graph Usage¶
To recreate the execution graph from SER files:
Read
pipeline_startrecords (record_type='pipeline_start') to obtain the canonical graph frompipeline_spec_canonical.For each SER record, use
identity.node_idanddependencies.upstreamto build edges.Combine with per-step information from
processor,context_deltaandassertions.
The canonical graph exposes processor_ref (FQCN); SER mirrors this as processor.ref.