stof

yer of data and communication between apps tends to be a bit messy, and strict adapters compiled at build time tend to be insufficient for the dynamism of modern software. This applies directly to AI-generated data and code, which is basically like calling into another untrusted service. With this, you can safely embed the execution, pick where things run, and combine harnesses like data. Works well as a "living" context, also with the JSON interop - just parse into or stringify sections of the Stof doc at runtime when needed, like agent tool defs. The playground on the site runs entirely in your browser. Stof is written in Rust and has support for JS/TS and Python today. I'd especially like feedback on whether the idea makes sense from the landing page, and where the docs need help. GitHub: https://github.com/dev-formata-io/stof Thank you!! -CJ

's a quick example: const users: [ { name: "Joe Schmo", height: 6ft + 1in, // Unit calculations built-in age: +32 } ] fn findUser(query: str) -> obj { for (const user in self.users) { if (user.name.lower().contains(query)) return user; } null } #[main] fn main() { const joe = self.findUser("joe"); let output = stringify("toml", joe); pln(output); } Use cases I'm excited about: - Configuration files that can validate and transform themselves - Data pipelines without separate processing code - API responses that contain their own transformation logic - Embedded systems running user-provided data safely The project is open source and looking for contributors. We're building something that will change how we think about the data/code boundary. Try it out: https://docs.stof.dev/ GitHub: https://github.com/dev-formata-io/stof What do you think? Has anyone else felt frustrated by the limitations of static data formats?