Create LarkBase fields automatically with n8n
A schema-safe workflow for comparing LarkBase fields, mapping types and creating only missing fields.
TL;DR: Field synchronization is a schema migration, not a blind create-fields loop. Read the current schema, compare stable field names or keys, infer types deliberately, then create only what is missing.
A four-step workflow
- List the destination fields and capture their IDs, names and types.
- Compare them with the intended schema using a stable key, not display order.
- Infer or map supported types explicitly, including select options and required configuration.
- Create missing fields only after a human-readable diff is approved.
LarkBase field type codes and option payloads should be kept in a versioned mapping in the workflow. A text field, number field, date field and single-select field do not have interchangeable configuration.
Treat changes as migrations
Never rename or change a field type silently when existing records depend on it. Keep manual columns out of generated ownership, validate API responses, and make the create step idempotent so a retry does not produce duplicates.
Before production
Run against an empty table and a table with existing fields, including a changed option set and an unsupported type. Log the before/after schema, stop on ambiguous matches, and document who approves future schema changes.