UltraciteUltracite
Dashboard

MCP

How to use Ultracite with MCP

The Model Context Protocol (MCP) is an open standard that allows AI assistants like Claude, Cursor, and other AI-powered tools to securely connect with external data sources and systems.

Ultracite provides an MCP server that gives AI assistants access to high-level coding standards and best practices — the same rules that guide AI code generation through Ultracite's editor integrations.

What It Provides

The Ultracite MCP server exposes a single tool:

  • getRules — Returns Ultracite's core coding standards covering type safety, modern JavaScript/TypeScript patterns, React/JSX best practices, error handling, security, performance, and testing guidelines.

These are high-level coding principles, not the specific linting rules from each provider (Biome, ESLint, Oxlint). For provider-specific rule details, see the Providers documentation.

Installation

Step 1: Choose Your AI Tool

Make sure you're using an AI development tool that supports MCP:

Step 2: Locate Your Configuration File

Depending on your AI tool, edit one of these files:

  • Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows)
  • Cursor: .cursor/mcp.json
  • Windsurf: .codeium/windsurf/mcp_config.json

Step 3: Add Ultracite Configuration

Add this to your MCP config file:

{
  "mcpServers": {
    "ultracite": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://www.ultracite.ai/api/mcp/mcp"]
    }
  }
}

Step 4: Restart Your AI Tool

Close and reopen your AI application for the changes to take effect.

Usage

Once configured, you can ask your AI assistant about Ultracite's coding standards:

What are Ultracite's coding standards?

The AI will retrieve the high-level principles covering:

  • Type safety and explicitness
  • Modern JavaScript/TypeScript patterns
  • React and JSX best practices
  • Async and Promise handling
  • Error handling and debugging
  • Code organization
  • Security considerations
  • Performance optimizations
  • Testing guidelines

Multiple MCP Servers

You can use Ultracite alongside other MCP servers:

{
  "mcpServers": {
    "ultracite": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://www.ultracite.ai/api/mcp/mcp"]
    },
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"]
    }
  }
}

Security and Privacy

  • The Ultracite MCP server only provides public coding standards documentation
  • No personal data or code is transmitted to our servers
  • No authentication required

How is this guide?

On this page