Getting Started

Litmus

Contract-driven test data generation for supply chain pipelines.

Define your dataset schemas as contracts, set referential integrity and coverage rules as constraints, group them into projects, and generate realistic synthetic data that satisfies every rule.

Key Concepts

Contracts

Schema definitions for each dataset — columns, data types, allowed values, and per-column constraints. Think of a contract as the "shape" your data must take.

Constraints

Rules between contracts — referential integrity (subset), coverage guarantees, cross-product relationships, and dimension matching. They make generated data realistic and self-consistent.

Projects

Group contracts and constraints together, configure row counts per table, and generate synthetic data. Download the output as CSV, Parquet, or both.

How It Works

Define or import contracts

Create schemas manually or bulk-import from a JSON schema file or CSV ontology export.

Create constraints between them

Define rules like "every arc must reference a valid node" or "inventory covers all product-location combinations."

Create a project

Group the contracts and constraints you need, configure row counts per table.

Generate test data

Click generate — Litmus builds data that satisfies every constraint, then verifies the results.

Download and use

Inspect generated tables, review constraint verification, and download a ZIP with CSV/Parquet files.

Try It Out

Litmus comes pre-loaded with a sample tenant containing a supply chain network model. Follow these walkthroughs to see everything in action.

1. Explore sample projects

2. Generate and download test data

3. Import your own ontology

4. Constraints and cross-validation

Constraints are the core of what makes Litmus-generated data realistic. Here's what you need to know.

Supported constraint types

Type	What it does	Example
`subset`	Every child value must exist in the parent column	Every `arcs.FROM_NODE` must be a valid `nodes.NODE_ID`
`coverage`	Every parent value appears at least once in the child	Every product has at least one demand record
`exactCoverage`	Child has exactly one row per parent value	One supply capacity row per node
`crossProduct`	Child has one row per combination of two parent columns	Inventory for every product-location pair
`dimensionMatch`	Both tables have the same number of rows	Header and detail tables must be 1:1 aligned

Cross-validation at author time: When you create or edit a constraint, Litmus validates it immediately. If rules are inconsistent, reference missing columns, or conflict with other constraints in the same project, you'll see errors before you ever generate data — not after.

Try opening arc-endpoints-valid in the sample tenant to see a subset constraint in action, or inventory-integrity for a cross-product example.

Ready to go?

Switch to the sample tenant to explore, or create a new tenant and import your own schema.