A2UI is an open standard and set of libraries that allows agents to generate rich, interactive user interfaces by sending declarative JSON descriptions that client applications render using their own native component libraries.
The problem A2UI addresses is that generative AI agents struggle to present rich, interactive interfaces to users, particularly when those agents are remote or operate across trust boundaries. Rather than having agents generate executable code, A2UI uses a declarative data format where agents describe the intent of a UI and client applications render it using pre-approved components from a trusted catalog. This approach treats agent-generated UIs as safe data while maintaining expressiveness. The format is designed to be LLM-friendly, allowing agents to generate and incrementally update UIs as conversations progress, with changes represented as modifications to a flat list of components with ID references.
A2UI suits projects where you need to let remote agents or language models generate user interfaces while maintaining strict security boundaries. It works well for applications built on any framework—web components, Flutter, Angular, React, SwiftUI—since the same JSON payload can be rendered across different platforms. The project emphasizes that developers retain full control over which components agents can request, and it supports custom component registration through a "Smart Wrapper" pattern, enabling developers to integrate existing UI components and enforce their own sandboxing policies. The tool is in early stage public preview with a stable v0.9 protocol family and a v1.0 specification in release candidate status, so the specification and implementations are still evolving.
The project's issue tracker is primarily used for internal planning rather than tracking inbound user demand, with most issues raised by the project's own members. Maintainers respond to new issues and pull requests within hours. Work in the tracker is dominated by feature and enhancement requests marked as sprint-ready, indicating active development focused on expanding capabilities.
