A UUID string identifying this external data schema.

Project ID of the project you're trying to access. To find the ID of the project, make a call to /api/projects/.

The latest error that occurred when syncing this schema.

Sync strategy: incremental, full_refresh, append, cdc, or xmin. * `full_refresh` - full_refresh * `incremental` - incremental * `append` - append * `webhook` - webhook * `cdc` - cdc * `xmin` - xmin

Column name used to track sync progress.

Data type of the incremental field. * `integer` - integer * `numeric` - numeric * `datetime` - datetime * `date` - date * `timestamp` - timestamp * `objectid` - objectid * `xid` - xid

Seconds to subtract from the stored incremental watermark at sync time, so each incremental run re-reads a rolling overlap window and catches late or backdated rows. Applies to timestamp/date incremental fields only. The stored watermark is unchanged. Maximum 5184000 (60 days).

How often to sync. * `never` - never * `1min` - 1min * `5min` - 5min * `15min` - 15min * `30min` - 30min * `1hour` - 1hour * `6hour` - 6hour * `12hour` - 12hour * `24hour` - 24hour * `7day` - 7day * `30day` - 30day

UTC time of day to run the sync (HH:MM:SS).

Column names for primary key deduplication.

For CDC syncs: consolidated, cdc_only, or both. * `consolidated` - consolidated * `cdc_only` - cdc_only * `both` - both

Names of source columns to sync. `null` (default) syncs all columns. Primary-key columns and the active incremental field are always retained, even if not listed here.

Predicates ANDed onto the source query so only matching rows sync. Each is `{column, operator, value}`; `null`/empty (default) syncs all rows. The operator must be one of `> >= < <= = != IN "NOT IN"` and the value must match the column's type (for `IN`/`NOT IN`, a comma-separated list like `1, 2, 3` or `'a','b'`). Applied on the next sync — not retroactive to already-synced rows.

Column metadata (name, data type, nullable) for this schema. For SQL sources this is the source-side schema discovered via `refresh_schemas`; for other sources (and once synced) it falls back to the synced table's columns. Empty only before the first successful sync/refresh.

Lightweight parent-source summary (id, source_type, column-selection support, the requesting user's access level). Only populated on the single-schema retrieve endpoint — `null` elsewhere — so read-only views can render without fetching the full source and all its schemas.