Skip to main content
In active development

The AI-native design system for React and Angular

AI-native components and blocks, available through an MCP server and a registry. Its documentation runs as a chain — design principles into components, components into the blocks built from them — so an LLM reads the whole system.

Designers prototype with the real components through MCP; developers ship the same ones to production.

85 primitivesReact + Angular editionsMIT licensed

Built for humans and language models

Four properties that make Gremorie different from a classical component library.

AI-native
Knowledge layer plus MCP serves the corpus to LLMs. Your design system is legible to language models, not just designers.
Two editions
Angular and React from the same registry. Same primitives, same semantics, same tokens. Shared corpus across stacks.
Copy-paste
Run gremorie add and the code is yours. No black-box dependency. Own every line, refactor without asking.
MCP-ready
Claude, Cursor, and Codex generate idiomatic UI by reading the MCP server. They build with your DS, not against it.

Generative UI artifacts

Render structured LLM output as production UI. Schema in, components out. Streaming-aware, type-safe, swappable.

Chart artifact
Monthly revenue, last 12 months
JSON SchemaStreaming
Code artifact
Generated component, ts + tsx
TSXRuntime

Theme everything with tokens

Three layers: primitive, semantic, chart. Override at any layer to retheme your entire app. Try a few changes below.

Customize tokens

Controls

Violet

Border radius8px
Font scale1.00x

Project Phoenix

Active

Tokens propagate from primitive through semantic and into every primitive. Change one variable, restyle the system.

Badge
Scoped preview. Try the sliders above.
85
primitives
8
categories
25+
corpus articles
  • AI-nativeKnowledge layer
  • MCP serverTool included
  • MIT licensedOpen source
Built by AI engineers, for AI engineers

Stop fighting your design system.
Make it AI-legible.

Open source. MCP-ready. Production-grade. Ship a brand-aware UI your LLM can read and your team can own in their next sprint.