Introduce Kanayago LSP that provides real-time Ruby syntax checking

[S-H-GAMELINKS]

Thank you for developing the Zed Ruby extension.

This PR adds support for Kanayago (金屋子), a Ruby Language Server that provides real-time Ruby syntax checking.
This is my first contribution to the Zed project, so please let me know if there is anything I should change or improve.

Summary

This PR adds the Kanayago extension to the Zed Ruby extensions repository.

What is Kanayago?

Kanayago is a Ruby Language Server Protocol (LSP) implementation that provides:

• Real-time Ruby syntax checking based on Ruby’s parse.y AST
• AST-based Ruby code analysis
• A lightweight LSP implementation written entirely in Ruby

Requirements

Users need to have the kanayago command available in their PATH:

gem install kanayago

Demo

zed-ruby-extension-kanayago-demo.mp4

Repository

• Kanayago LSP: https://github.com/S-H-GAMELINKS/kanayago

[andyw8]

I'm curious, what are the use cases for choosing this over Prism?

[S-H-GAMELINKS]

Thanks for your comment.

Prism is a hand-written recursive descent parser. This is a very deliberate design choice and provides a lot of flexibility for tooling. As a trade-off, supporting new Ruby syntax usually requires explicit manual updates to the parser.

Kanayago, on the other hand, works with the AST generated directly from Ruby’s parse.y. Because it follows CRuby’s official grammar, support for new syntax tends to come more naturally as Ruby itself evolves, without requiring a separate parser implementation to be updated.

This leads to a few practical use cases where Kanayago is especially useful:

• Trying out or validating new Ruby syntax in the editor
  Kanayago runs CI against Ruby’s master branch. While this is not an official support guarantee, it provides a practical environment for experimenting with upcoming or recently introduced syntax directly in Zed.

• Early, strict syntax feedback
  Because parse.y is an LR parser, Kanayago inherits stricter grammar validation from CRuby itself. This makes it well-suited for catching syntax errors early, even in incomplete or experimental code.

There’s also a subtle edge-case difference worth mentioning. Because Prism is hand-written, there have been rare cases where syntactically invalid Ruby code was accepted as valid. One such real-world example is documented in a Ruby core issue:

https://bugs.ruby-lang.org/issues/21337

Since Kanayago relies on parse.y, these kinds of edge cases are much less likely.

[janko]

The Ruby LSP + Prism combo already provides everything you mentioned. Since Prism is now the default Ruby parser (not parse.y), it's going to be updated with new ruby syntax. Bugs can still happen of course, but the same is true for parse.y. I believe the goal of Prism was for Ruby to eventually be able to move away from parse.y.