The Model Context Protocol (MCP) is an open standard that allows AI coding assistants and LLMs to interact securely with external tools, APIs, and data sources. With SCHMA's official MCP server, your AI editor can inspect content models, create entries, synthesize schemas from natural language prompts, and manage published content directly from your coding workflow.

What is the Model Context Protocol in SCHMA?

Rather than copying and pasting schema definitions or toggling between your code editor and a browser tab, SCHMA provides @schmadev/mcp-server. When connected, your AI assistant can execute validated actions against your workspace through structured MCP tool calls.

Prerequisites and credentials

Before configuring your editor, ensure you have gathered three required configuration values from your SCHMA Studio workspace:

• SCHMA API URL (`SCHMA_API_URL`): https://api.schma.dev (or your custom API gateway endpoint).
• Management API Key (`SCHMA_API_KEY`): A token generated in Settings → API Keys with read_write permissions scope.
• Workspace Slug (`SCHMA_TENANT_SLUG`): Your workspace's unique slug identifier.

Configuring Cursor

Cursor supports MCP servers natively. Open Cursor Settings → Features → MCP, or edit your configuration file directly (~/.cursor/mcp.json or .cursor/mcp.json in your project root):

{
  "mcpServers": {
    "schma": {
      "command": "npx",
      "args": ["-y", "@schmadev/mcp-server"],
      "env": {
        "SCHMA_API_URL": "https://api.schma.dev",
        "SCHMA_API_KEY": "schma_rw_your_management_token_here",
        "SCHMA_TENANT_SLUG": "your-workspace-slug"
      }
    }
  }
}

Configuring Claude Desktop

In Claude Desktop, open Settings → Developer → Edit Config (located at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS or %APPDATA%\Claude\claude_desktop_config.json on Windows):

{
  "mcpServers": {
    "schma": {
      "command": "npx",
      "args": ["-y", "@schmadev/mcp-server"],
      "env": {
        "SCHMA_API_URL": "https://api.schma.dev",
        "SCHMA_API_KEY": "schma_rw_your_management_token_here",
        "SCHMA_TENANT_SLUG": "your-workspace-slug"
      }
    }
  }
}

Available MCP tools and capabilities

Once connected, restart your editor's MCP server. Your assistant gains access to the following verified tools:

• list_content_types: Inspect all schemas and content models in the workspace.
• get_schema: Retrieve complete field configurations and validation rules for a content type.
• get_entries: Query published or draft entries by content type slug with pagination.
• create_entry: Draft new entries with field values validated against the target schema.
• update_entry: Modify existing content entries by entry ID.
• publish_entry: Transition draft entries to published status.
• create_content_type: Programmatically create a new content type schema.
• generate_schema: Use AI to synthesize a complete content model from a natural language prompt.