zod-to-json-schema VS typebox

Zod and GPT-4 Zod is great for generating typed JSON from GPT-4, because you can easily convert Zod schemas to JSON schemas using zod-to-json-schema and include descriptions for each property. The JSON schemas are passed to GPT-4 as function definitions or in the prompt (when you are using JSON mode).

https://github.com/StefanTerdell/zod-to-json-schema#readme

The documentation can be found here.

So far I like the commonly used approach in the Typescript community best:

1. Create your Schema using https://zod.dev or https://github.com/sinclairzx81/typebox

2. Generate your Types from the schema. It's very simple to create partial or composite types, e.g. UpdateModel, InsertModels, Arrays of them, etc.

3. Most modern Frameworks have first class support for validation, like is a great example Fastify (with typebox). Just reuse your schema definition.

That is very easy, obvious and effective.

There's a ton of schema validators out there and most devs have their personal favorite. Mine was zod and is now typebox + ajv.

If you're willing to document your API with an OpenAPI schema, then it should be possible to generate TypeScript types based on the OpenAPI schema with something like openapi-typescript. Also, Typebox can generate JSON schemas, maybe it can be used to generate something that the front-end can also use?

Well, I do provide extensibility for those bullet points you've listed to varying degrees (both schema and type representation), as well as offering a reference implementation for expressing a entirely different schema specification under the type system (specifically RFC8927 / JSON Type Definition). Reference implementation here. As for JSDoc, It's supported in code hints.

In setting out to build this service, we wanted to use gRPC for its APIs. We’ve been reaching for REST when building APIs so far, primarily out of necessity, i.e., our public APIs needed auto-generated client SDKs and docs for developers working with them. We built those APIs with Fastify and Typebox but felt burned by a code-first approach to generating an OpenAPI spec. I’ll spare you the details and save that experience/learning for another article. Suffice it to say we love gRPC’s schema-first approach. This blog post summarizes our feelings well