What is the synthetic data corpus, and how do I use it?

What it is

The corpus is a public GitHub repository, Trazomo/Trazomo-Synthetic-Data: matter files, contracts, correspondence, ledgers, and datasets for a fictional law firm, Atticus Dundee LLP, its clients, and the companies around them. Everything in it is invented. Any resemblance to real people or companies is coincidental.

That is the point. You practise on data that looks like your work without ever putting client material into an AI tool.

The universe and its clock

Every path works inside the same small world, so a document you meet in one module can reappear in another and still make sense.

Who	What they are	Which paths meet them
Atticus Dundee LLP	The law firm. Its matters, clients, and correspondence are the legal track's home ground.	Legal
Atticus Dundee Inc.	The firm's flagship client: a roughly 600-person B2B software company selling a workflow and collaboration product. Its books, people, pipeline, delivery teams, and policy library are where the applied work happens.	Finance, Revenue, People and HR, Operations, and Legal from the in-house seat
Larkspur Design & Build	A small design-and-build business that buys from and works alongside the same cast.	Small Business

The universe has one clock. "Now" is the first week of April 2026. The master services agreement between the software company and its vendor was signed on 2026-01-12 with a subscription start of 2026-02-01, and the finance datasets cover March 2026: the bank statement and cash ledger run 2026-03-01 to 2026-03-31 and the month-end close happens in the first week of April. When a lesson asks you to decide something, decide it as someone standing on that date with the information the universe had then.

The people are fictional too. Names, roles, and reporting lines live in the corpus canon so that a preparer, a reviewer, or a client contact is the same person from one artifact to the next.

How lessons use it

When a lesson hands you a document, it names the artifact and the version, for example "LGL-02 · Trazomo-Synthetic-Data v1.0.1", and links the file. Use the version the lesson names so your results match the lesson. Answer keys stay inside the practicum; the repository never contains them.

Getting the files

• Clone the repository, or open a file on GitHub and use its Raw view.
• Point your coding agent at the repository. It is written agent-first: AGENTS.md and MANIFEST.json explain the layout, and artifacts/, datasets/, canon/, and specs/ hold the material.

Licence

The generator code is MIT. The dataset and canon content are dedicated to the public domain (CC0), so you can copy, adapt, and share them freely.