Canvas data is included in Space export and import, whether or not Canvas is available in the Space.
Path | Contents |
|---|---|
| Canvases, versions, audit records, provenance records, candidates |
| System capabilities, system Flow definitions, conversion records |
Both files are optional on import. A missing entities/canvases.json is treated as { canvases: [] }; a missing entities/canvasSystem.json is treated as { capabilities: [], flowDefinitions: [], conversions: [] }. A file that fails validation rejects the import with Invalid Canvas data.
Export includes one entry per active or archived Canvas in the Space. Canvases that have been deleted are excluded.
Each entry contains:
Field | Contents |
|---|---|
Canvas fields |
|
|
|
|
|
| All provenance fields except |
| Candidates, each with its ordered |
System capabilities, system-owned Flow definitions, and conversion records are exported alongside the Canvases.
Import rewrites every ID that is scoped to the Space, so imported content refers to the destination Space's IDs.
Snapshot references are rewritten and the snapshot is re-canonicalized:
Node data | Rewritten |
|---|---|
Any node data carrying |
|
|
|
|
|
|
|
|
|
|
|
If a referenced ID has no mapping in the import, the import fails with Missing imported <type> mapping for <id>.
System capabilities, system Flow definitions, and conversion records are imported as well.
Imported provenance records are marked as severed. Their run details — producerDefinitionId, runId, scopeId, activationId, attemptId — do not refer to runs in the destination Space and are replaced by the severed marker.
Author, time, artifact IDs, and artifact hashes are preserved. Candidate decisions and their acceptance records are kept, so content accepted in the source Space remains accepted after import.
Imported content cannot be re-run. See Canvas Review and Provenance.