microsoft/typespec

TypeSpec is a language from Microsoft for describing the shape of cloud service APIs. Instead of writing an API description once in one format and then separately maintaining documentation, client code, and server stubs, you write a single TypeSpec definition and use it to generate all of those outputs automatically. The idea is that one source file stays authoritative while everything derived from it stays in sync. APIs today come in many styles: REST endpoints described by OpenAPI documents, gRPC services described by protocol buffer files, and others. TypeSpec is designed to cover all of those shapes with a common vocabulary. From one TypeSpec file you can emit an OpenAPI 3.0 document, a gRPC definition, or other formats depending on which output libraries you bring in. The language includes a way to package reusable patterns into libraries. A team can define standard conventions once and share them across many API projects, so every new API starts from an agreed baseline rather than a blank slate. There is also a linter framework that can flag when code drifts from those conventions. Practically, you install the TypeSpec compiler via the npm package manager, write your API definition in a .tsp file, and run a compile command that produces the output format you need. Extensions are available for VS Code and Visual Studio to provide editing support while you write TypeSpec files. The project is maintained by Microsoft but described as community-supported, meaning external contributions are accepted. It is aimed at teams building or documenting APIs who want a single definition to be the source of truth rather than keeping multiple format-specific files manually in sync.