kazeflow runs

Read portable records from an existing local SQLite store. History commands never create, initialize, migrate, or write the database.

List

kazeflow runs list [--store PATH] [--limit N] [--format text|json]

Records are ordered by saved time and run ID. The default read path is ./.kazeflow/runs.sqlite3.

Show

kazeflow runs show RUN_ID [--store PATH] [--format text|json]

show returns the stored portable envelope. Unknown IDs are selection errors.

Compare

kazeflow runs compare LEFT_RUN_ID RIGHT_RUN_ID \
  [--store PATH] [--format text|json]

Comparison preserves left/right argument order and compares run and task aggregates. It does not match dedicated raw partition-key fields because those fields are not stored. Failure messages and tracebacks remain application-controlled and can repeat the same values; stored records are not automatically redacted. See the portable-record trust boundary.

For the typed JSON envelopes, distinct CLI/record/store versions, schemas, and ordering guarantees, use the JSON automation contract.

History JSON uses schema version 4. Each listed or shown run reports its own portable record version, so a store schema v1 database can return a mixture of record v1 through v4 rows without rewriting older runs. New saves use record v4, which includes successful resolution and materialization IDs. Viewing history only displays the saved metadata; it does not contact the adapter or load artifacts.

$ kazeflow runs list
Stored runs:
- <run-id> (success; saved_at: <timestamp>; schema: 1)

$ kazeflow runs show <run-id>
Stored run:
- run_id: <run-id>
- status: success
- saved_at: <timestamp>
Portable record:
{ ... }