Templates

A template is a named list of fields used to generate data. It belongs to a project and can be reused from the editor, chat, scenarios or API.

Define fields

Each field has a name and a generator type. Options depend on that type:

{
  "name": "Customer",
  "fields": [
    {"name":"id","type":"UUID"},
    {"name":"first_name","type":"First Name"},
    {"name":"last_name","type":"Last Name"},
    {"name":"email","type":"Derived","options":{"value":"{{first_name}}.{{last_name}}@example.com"}}
  ]
}

Use names from the field catalogue. Generation quantity is supplied when you generate; the REST API uses POST /datamaker with fields and quantity.

Create, inspect and reuse

1. Create a template in the editor or ask the agent to propose one from sample data or a connected schema.
2. Review field types and options. Inferred fields are suggestions, especially when the input sample is small.
3. Generate a few rows and inspect formats, nulls and relationships.
4. Save the template and use its ID from your scenario or SDK.

Template updates change the stored definition. Do not assume every save creates a recoverable revision or that an @v7 ID pins an older version. Keep reviewed field definitions in source control when you need reproducible fixtures.

Templates can be grouped into folders once a project holds more than a screenful, and saved endpoints group the same way in Connections. A folder is organisation only: it carries no permissions and does not scope generation.

Control the result

• Distributions defines weighted values and seeded generation.
• Nested objects composes inline structures and references other templates.
• Python Script fields runs custom script output when ordinary generators are insufficient.
• Sensitive fields explains classification and actual masking controls.
• Schema discovery and custom types reuses source structure and value libraries.

Internal generators support a request seed. External AI, API, database and Python values can change independently of that seed. Inspect generated fields for error markers before treating a response as a valid test fixture.