Databases

Structured databases hold typed records alongside your files. Use them for anything an agent should query rather than read: project trackers, glossaries, decision logs, content calendars.

Schema

A database has typed properties. Supported types are text, number, boolean, date, select, multi_select, url, relation, file_reference, and asset_reference.

Record values are keyed by stable property UUID rather than property name, so renaming a property never breaks API consumers:

{
  "values": {
    "22222222-2222-4222-8222-222222222222": "Launch",
    "33333333-3333-4333-8333-333333333333": 42,
    "44444444-4444-4444-8444-444444444444": true
  }
}

References and page bodies

• A file_reference is one file UUID or, when the property enables multiple files, an array of UUIDs.
• An asset_reference identifies an imported binary attachment.
• A record can set contentFileId to use a normal Orcha file as its editable page body.

Referenced resources must be active and in the same workspace. Folder-scoped tokens cannot write out-of-scope references, and inaccessible references are redacted from responses.

Views, versions, and retrieval

Databases have saved table views with filters and sorting. Updating a record merges values and saves a version. Records are indexed automatically: search_context results include native record citations linking to /databases/:databaseId?record=:recordId.

Query records with search, sorting, filtering, and pagination (limit up to 200). Filter operators are contains, equals, not_equals, is_empty, and is_not_empty. See the API reference and the MCP database tools.