What problem does it solve? Moving records in and out of a Salesforce org is error-prone: loads that cannot be re-run create duplicates, children load before parents, bulk jobs fire and forget, and destructive schema changes silently lose data. This Skill provides the decision rules, command syntax, limits, and recipes to seed, migrate, backfill, extract, and delete Salesforce data safely. ## Core Features & Use Cases - Mechanism selection by volume: sObject tree for hand-written seed data, single-record commands for ad hoc changes, Bulk API 2.0 for anything over 2,000 records, and bulk export for queries over 10,000 rows. - Repeatable loads with external ids: upsert patterns keyed on unique external id fields, relationship columns for parent-child loads in one pass, and load-ordering rules for tree plans. - Limits and safety planning: Bulk API 2.0 allocations (15,000 batches per 24 hours, 100 MB practical file size), hard delete and Recycle Bin behavior, and data-loss rules for destructive schema changes. - Use Case: Seed a scratch org by exporting a shaped Account-Contact sample with sf data export tree --plan, committing it, and replaying it with sf data import tree on every new org, then verifying counts and orphan Contacts. ## Quick Start Ask the assistant to seed your scratch org with the committed data plan, or to write a repeatable bulk upsert script for a CSV of Accounts keyed on an external id field.
