Documentation is a hated task among developers, but being out of sync with code will lose its purpose. A solution for this issue is to let machines handle it, freeing up developers' time for more critical tasks. With the advent of GitHub's free Copilot feature, one could generate some automated documentation, but the result may not be what you want, as it cannot handle more complex imperative code well. ZenStack's plugin system, built on top of Prisma, solves this issue by enhancing the schema access policies and validation rules and can be considered the single source of truth for the back-end application of your project. ZenStack also automatically generates APIs and corresponding front-end hooks for you.
Focusing on the documentation generation problem, ZenStack has leveraged AI with the help of Vercel SDK, which provides a standardized approach to interact with various Language Model Modelling (LLM) providers, allowing you to write code that works with multiple AI models, such as OpenAI or Anthropics Claude. The Zen-Stack-markdown package offers a plugin-based architecture for generating a markdown for a ZModel schema, implementing features around it as plugins, allowing you to create technical design documents directly from the ZModel schema file. Zen-Stack-mark-down uses a unified API allowing you to specify any LLM model you wish to use and generates ERD diagrams for each model. This allows you to create type-safe documentation for cases where copilot documentation falls short. This means developers can save time while avoiding the cumbersome task of maintaining documentation.