Database Metadata
Examples: Table names, column names, data types, keys, relationships, and schema structure.
Purpose: Used to understand database structure and provide context for schema-aware SQL generation.
Trust
How SQL Mocker Uses Your Data
SQL Mocker uses database structure and query context to help generate and review SQL. This page explains the different types of information involved and how they are used.
Information by category
Natural-Language Questions
Examples: For example: ‘Show active construction projects by contractor.’
Purpose: Used with the selected schema context to determine the intended result and generate SQL.
Existing SQL
Examples: SQL that a user pastes or uploads to the Imported SQL workspace.
Purpose: Processed to explain, review, update, optimize, identify joins, or check query structure.
Dummy Data
Examples: Generated sample values based on the reviewed schema and selected locale.
Purpose: Used for previews and query checks inside the workspace. It is generated data, not a copy of production rows.
Production Database Rows
Examples: Customer records, transactions, operational notes, and other values stored in real tables.
Purpose: Not required by schema-based SQL generation. Users should avoid manually submitting private or sensitive rows unless authorised to do so.
Schema-based SQL generation flow
You choose and review the context that enters the workspace. SQL Mocker sends the relevant schema context and request to the AI, then returns SQL for you to inspect.
- Your database
- Schema metadata
- SQL Mocker workspace
- Relevant schema context + request sent to AI
- SQL generated
- You review and run it in your database environment
What SQL Mocker needs vs what it doesn't need
SQL Mocker may use
- Table and column names
- Data types
- Keys and relationships
- Existing SQL submitted for review
- Natural-language questions
Schema-based workflows do not require
- Production database rows
- AI access to the live database
- Database credentials supplied to the AI
These are workflow requirements, not a technical block on manual input. Review prompts, files, SQL, and support messages so you do not submit credentials or sensitive records.