1. Source attribution
Every record carries provenance that records where it came from and how it was created. The lineage system tracks:- Origin: Where the record originated (
origin_feature,origin_source_id,source_details). - Transforms: Each transform that produced or enriched the record, including
transform_name,connector_name, andexecuted_by. This links each piece of data to the connector that produced it.
GET /api/provenance/{id}/lineage for a single record’s origin, transforms, transitions, and updates. Use GET /api/provenance/{id}/full-lineage for the complete ancestry chain from root to current record, including a unified timeline of all events.
2. Lineage chain and integrity
The full lineage response includes:chain: Ordered list of records from root (oldest ancestor) to current. Each step contains its events (origin, transforms, transitions, updates, deduplication, merges) in chronological order.timeline: Unified list of all events across the chain, sorted by timestamp.merge_history: Records of merge operations that contributed to this entity.
3. Export provenance
PDF exports can include provenance metadata wheninclude_provenance is enabled. The export records:
| Field | Description |
|---|---|
exported_at | UTC timestamp when the export was generated |
exported_by | User ID who triggered the export |
export_config | The resolved export options |
data_fingerprints | SHA-256 hash of each section’s data at export time |
truncation_summary | Counts when data was truncated |
4. Audit and accountability
The lineage timeline attributes every event to a user or system:created_by/created_by_name: Who created the record (origin events)executed_by/executed_by_name: Who ran the transformupdated_by/updated_by_name: Who modified fields (update events)
field_path, old_value, new_value, and source. Transitions record when data moved between features with from_feature, to_feature, and trigger_type.
PDF export
Report PDF exports can include:- Report content
- Export provenance section (when
include_provenanceis true): exported_at, exported_by, export config - Dataset fingerprints: SHA-256 hashes for integrity verification
- Truncation summary when data was paginated