Raven
Give your coding agent structured design rules so generated landing pages, dashboards, and marketing UI stay consistent instead of looking like generic AI slop.
Overview
Raven is a Build-phase MCP server that supplies design intelligence—principles, patterns, content, brand, and tokens—for AI-generated UI in your agent workflow.
What is this MCP server?
- Design principles and UI patterns tuned for agent-generated interfaces
- Content, voice, and brand guidance agents can apply per screen
- Design token and styling conventions for cohesive multi-page products
- stdio npm package (raven-mcp v1.3.3) for local Claude Code / Cursor MCP config
- Remote catalog entry at ravenmcp.ai tied to github.com/rhinocap/raven-mcp
- Server version 1.3.3
- npm package identifier raven-mcp (stdio)
- Publisher site ravenmcp.ai
Community signal: 1 GitHub stars.
What problem does it solve?
Agent-built interfaces drift across pages because the model has no persistent design system or brand rules to follow.
Who is it for?
Indie SaaS founders using Claude Code or Cursor who want on-brand UI generation without rebuilding a design doc on every prompt.
Skip if: Teams that only need raw Figma-to-code handoff with no agent in the loop, or backends with no user-facing UI.
What do I get? / Deliverables
After you register Raven, your agent can reference consistent UI patterns and tokens so new screens match your product’s visual and content standards.
- Agent-accessible design principles, patterns, and token guidance for UI tasks
- More consistent AI-generated layouts and copy aligned to your brand
- Repeatable MCP connection via local stdio without custom API wiring
Recommended MCP Servers
Journey fit
UI is assembled in Build, but design intelligence is most valuable while you are still shaping screens and components. Frontend work is where principles, patterns, tokens, and brand constraints directly change what Claude or Cursor outputs in HTML/CSS and component libraries.
How it compares
Design-intelligence MCP server, not a component npm library or a standalone Figma plugin.
Common Questions / FAQ
Who is Raven MCP for?
Solo and indie builders who generate or refactor frontend UI with AI agents and want principles, patterns, and tokens enforced in context.
When should I use Raven MCP?
Use it during frontend builds and UI refactors whenever you want agent output to follow a defined brand and pattern library instead of one-off styling.
How do I add Raven MCP to my agent?
Install the raven-mcp npm package (stdio v1.3.3), add the server block to your host’s MCP config (Claude Desktop, Claude Code, Cursor, etc.), restart the client, and invoke design guidance from your UI prompts.