What Scratch edits in PostgreSQL
- Text and varchar columns in any table
- JSON and JSONB columns
- Numeric, boolean, date, and timestamp columns
- Row-level edits without schema changes
How it works
Scratch pulls a table from your Postgres database. Each row becomes a local file. Your AI edits the columns you allow. Primary keys, generated and identity columns, and any column the database reports as non-updatable stay locked, so serial IDs and computed values are never overwritten. Column types, length limits, and foreign keys travel with the schema, and the database rejects any write that would violate a constraint. You approve in Scratch. Scratch updates the rows through the standard Postgres driver.
Connect with a single connection string. Scratch discovers tables across every non-system schema, and a table needs a primary key to sync. For incremental pulls, point Scratch at a timestamp column like updated_at. Best for content tables — posts, product copy, knowledge base — that live in Postgres without a CMS in front of them.