Much like how servers used to render HTML, LMUI is a protocol that guides LLM apps/clients to render specified components.
I'm not sure if anyone's already built an existing protocol, but I really think pure text is killing the potential of LLM clients, especially with how much more interactivity MCP's have brought for us to play with existing services. Improving the UI/UX should be a next step in LLM applications.
PR/feedback is very welcome.
Show HN: A protocol for LLM client to render UI components | Heykuki News