Roadmap

What is next

Grit ships every merged change as a release, so the changelog says what happened and says it in a lot of detail. This says what has not happened yet, which is the other half of deciding whether to build on something.

Reviewed 2026-11-22. Every line links to its issue, which is where the real state is: this page is a statement of intent, and intent is the part that only a person can refresh. Nothing here is a date, because a one-maintainer project that promised dates would be lying on a schedule.

Next

6

Started, or the next thing to start.

Later

29

Agreed it should exist, not scheduled. Say so on the issue if you need one sooner: that is how several of these moved up.

Waiting on a decision

1

Open because a decision is owed before any code is worth writing. The decision is named, so it can be argued with.

Not planned

Decisions already made, with the reason, so you can disagree with the reason rather than guess at one. A request nobody has got to yet is in the list above, not here.

  • A second web framework, ORM or CSS library

    Gin, GORM and Tailwind are the stack, and the value of a convention-led framework is that there is one answer. Supporting Echo as well as Gin would double the generator and halve the conventions.

  • A Pages Router option for the Next apps

    App Router only. Every generated page, layout and server action assumes it, and the second shape would be the one nobody tests.

  • Radix as the primitive library

    Where a new primitive genuinely needs behaviour rather than appearance, it uses Base UI, which is the successor the shadcn maintainers point at. Existing hand-written primitives stay: they work and they are tested, and rewriting them would trade known behaviour for unknown.

  • A component registry inside every scaffolded project

    It was there and was removed in v3.31.78. Every project carried a UIComponent model, a registry handler, a 91-component seeder and a browser page, for a feature almost nobody used. Grit UI is a standalone library instead.

  • An access column on the routes screen

    gin exposes a route’s last handler and nothing in front of it, so the middleware chain cannot be recovered at runtime. The static table in internal/access is how anything learns which routes are protected, and inventing the column from path shapes would be a guess presented as a fact.

Want one of these sooner

Say so on its issue, with what you are building and what the absence costs you. That is not a formality: several items moved up because one person explained the case, and three of the things on this month's changelog exist because somebody filed the issue rather than working around it.