MCP 服务器

gemini-designer-mcp

Provides AI-powered UI design, repo intelligence, and frontend code generation tools, enabling assistants to scaffold projects, generate and modify UI components, manage design tokens, and analyze repositories.

README

Gemini Designer MCP

AI-powered UI design and code generation for your coding assistant

An MCP server that gives your AI coding assistant 17 tools for UI design, repo intelligence, and frontend code generation.

Install

npx gemini-designer init --local

That's it. The CLI will:

Detect your MCP clients (Cursor, Gemini CLI, VS Code, Claude, Codex, Windsurf, Zed, Antigravity)
Let you choose which to configure
Set up Gemini Designer automatically

Tools

Tool	What it does
`generate_vibes`	Design tokens, themes, color palettes
`scaffold_project`	Plan + scaffold multi-file project structure (React/Next/Vue/SvelteKit/Remix/Angular)
`ui_generate`	Unified generator that returns a multi-file `FileBundle` (no writing)
`modify_ui`	Edit existing UI code (supports multi-file targets)
`generate_component_variants`	Add variants (+ optional Storybook file)
`ui_review`	Screenshot + target files → patchable `FileBundle`
`analyze_tokens`	Extract tokens from CSS
`sync_design_tokens`	Convert tokens between CSS/Tailwind/Tokens Studio
`detect_ui_stack`	Auto-detect framework & libraries
`catalog_components`	Index your components
`repo_tree`	Fast filtered repo file listing
`repo_read_files`	Read multiple files safely (truncation + sha256)
`repo_get_file_meta`	File metadata (exists/size/mtime + optional sha256)
`repo_search`	Text search (ripgrep-powered)
`apply_file_bundle`	Generic writer tool to apply multi-file bundles
`validate_project`	Run allowlisted scripts (lint/typecheck/build) (disabled by default)
`prompting_guide`	High-signal prompting templates

Usage

Local mode is free and uses your own Gemini API key:

npx gemini-designer init --local

The server also retains a remote mode for compatible self-hosted gateways. The former hosted Gemini Designer service is currently unavailable and is not required for local use.

Auto-Detected Clients

The CLI automatically detects and configures these MCP clients:

Client	Config Path
Cursor	`~/.cursor/mcp.json`
Gemini CLI	`~/.gemini/settings.json`
VS Code	`~/.vscode/settings.json`
Claude Code	`~/.claude/mcp.json`
OpenAI Codex	`~/.codex/config.toml`
Windsurf	`~/.codeium/windsurf/mcp_config.json`
Zed	`~/.config/zed/settings.json`
Antigravity	`~/.gemini/antigravity/mcp_config.json`