sf-data

Use sf data import bulk with a CSV file for large volumes, or run an anonymous Apex factory script via sf apex run for controlled generation. For trigger testing, create at least 251 records to cross the 200-record batch boundary.

Run sf sobject describe on the target object first and filter for fields where nillable is false and createable is true. Include those fields with valid values, and verify picklist entries against the describe output rather than guessing.

Use single-record sf data create or update commands for roughly 1-10 records, standard API patterns up to about 2000 records, and Bulk API 2.0 via sf data import bulk for anything larger. Bulk API supports up to 10 million records per 24 hours.

No, Database.rollback does not undo async operations like future, queueable, or batch jobs, and transactions allow a maximum of 5 savepoints. For async scenarios, use delete-by-pattern, delete-by-created-date, or ID-tracking cleanup instead.

Delete children before parents, use specific name prefixes like 'Test%' to avoid deleting real data, and query records first to preview what will be removed. For large volumes, export IDs to CSV and run sf data delete bulk.

Do not use it for pure SOQL query writing, Apex test execution, or metadata deployment, which belong to dedicated query, testing, and deployment skills. It also requires that target objects and fields already exist in the org before data creation.