Skip to content
DataMaker Docs

Concepts

The whole DataMaker model fits on one page. There are only six objects worth knowing.

Workspace

Your team. Has a name, a billing plan, members, and one or more projects. API keys are scoped to a workspace, but most actions happen inside a project.

Project

A logical grouping of work — typically one per environment (dev, staging, regression-suite) or one per system under test. Templates, connections, scenarios, chats, and assets all live inside a project.

You can move objects between projects in your workspace; you can’t share them across workspaces.

Template

A reusable schema definition: a list of fields, each with a type and options. Templates are the unit of generation. They’re versioned (every save creates a revision), they have an ID, and they can be invoked from the UI, the REST API, an SDK, the MCP server, or the chat agent.

Templates can nest: a User template can have an address field whose type is “another template”, which itself has fields like street, postcode, city. See Templates → Nested objects.

Connection

A configured target system — a database (Postgres / MySQL / MongoDB / MSSQL / Oracle / DB2), a REST endpoint, or an SAP OData service. A connection knows the URL, the auth method, and (for OData) the entity sets it discovered from $metadata.

Connections are how generated data leaves DataMaker. See Connections.

Scenario

A Python script that orchestrates a sequence of DataMaker actions — generate from template A, POST to API B, insert into DB C, validate, log. Scenarios run on DataMaker’s hosted Python workers, stream their logs back to the UI, and can be triggered from the REST API, the MCP server, the chat agent, or a CI pipeline.

Use scenarios when a single template-and-export step isn’t enough — for example when you need referential integrity across templates, or when you need to call a non-DataMaker API in the middle of a seeding run. See Scenarios.

Agent

DataMaker ships an AI chat agent (built on Anthropic’s Claude). The agent has access to ~25 MCP tools that wrap every meaningful action — list templates, generate, push to endpoint, create or run a scenario, fetch SAP $metadata, and so on. You drive it from plain-English prompts in the Chat view; the agent decides which tools to call.

Agent mode also exposes the same toolset to external AI agents (Claude Code, Cursor, Copilot) via the DataMaker MCP server. Same tools, your editor.

How they fit together

Workspace ── Members, billing
   │
   └─ Project ── Connections (Postgres, REST, SAP OData…)
          │
          ├─ Templates  ─── reusable schemas, versioned
          │
          ├─ Scenarios  ─── Python scripts orchestrating generation + push
          │
          └─ Chats      ─── conversations with the AI agent
                           (which calls MCP tools that act on
                            templates / scenarios / connections)

If you understand those six things, you understand DataMaker.

Where to go next