Exploring schema-safe mock data generation and improving existing tooling

[samriddhitiwary]

Hi everyone 👋

I wanted to start a discussion to get broader feedback on a tooling idea and understand whether it aligns with the project’s direction.

My idea is to focus on schema-safe mock data generation using existing tools like faker.js or similar generators, while strictly keeping the output schema-compliant. The motivation here is that many existing generators currently produce fairly basic or placeholder data, and there seems to be room to improve the usefulness of the generated data without compromising correctness.

Along with this, I also plan to explore adding an NLP-based layer to generate more meaningful values where the schema allows, while still ensuring all constraints are respected.

Instead of starting something entirely from scratch, I was wondering if it would make more sense to contribute to and enhance an existing tool in this space. I came across https://github.com/sagold/json-schema-library and wanted to ask whether enhancing that library (or a similar one) would be a better direction.

[gregsdennis]

There is also my library, JsonSchema.Net.DataGeneration. It needs a little work, and I need to publish a new version that works with the newest JsonSchema.Net.

It supports almost everything except things that need regex (e.g. pattern).