Chakra UI v3's snippet-based rewrite: migrate an existing v2 app or pin to v2 long-term?

The compatibility boundary

Chakra UI v3 is a ground-up rewrite on Ark UI and Panda CSS rather than an incremental release. The v2 Emotion runtime styling engine, the extendTheme/styleConfig theming API, and monolithic component imports are replaced by createSystem recipes, slot recipes, and CLI-generated snippet components composed from parts. There is no backward-compatibility layer, and running v2 and v3 side by side in one app is not a supported path.

The decision

A team maintaining a mid-sized v2 application with a custom theme and several dynamic, runtime-computed style props must choose between two expensive options: run the official codemod and manually rewrite what it cannot convert, or pin to v2 and accept that the line is no longer the focus of new development. Static extraction in v3 also means some v2 dynamic styling patterns may fail silently—producing missing styles rather than build errors—which complicates effort estimation.

Open questions

• For an app relying heavily on runtime-computed style props, is a full v3 migration realistic, or does the static-extraction model make pinning to v2 the safer long-term choice?
• How much of a typical v2 theme does the official codemod actually convert, and which APIs consistently require manual rewriting?
• Is there any documented incremental migration strategy, or is a single cutover the only viable path given that mixed v2/v3 usage is unsupported?