Base44 MCP Server

Base44 is an AI-powered no-code platform for building fully functional applications (internal tools, CRMs, dashboards, customer portals) from natural language. The official Base44 MCP server exposes your Base44 account to any MCP-compatible AI assistant such as Claude, ChatGPT, or Cursor, letting the assistant create new projects, modify existing ones, and inspect or query the data inside them.

The server is hosted by Base44 at https://app.base44.com/mcp and uses OAuth for authentication, so users sign in once through the standard Base44 OAuth flow rather than managing API keys. Once connected, the assistant can spin up apps from a description, list the apps in your account, inspect entity schemas, and run queries against live data. Builds run asynchronously, and the server returns preview URLs when projects are ready.

Because the server is officially maintained by Base44, it stays in sync with the platform and benefits from vendor support. Note that there are also community implementations on GitHub (such as maorfr/base44-mcp) and an unofficial SDK-docs MCP, but those are not maintained by Base44 itself.

Tools

Prerequisites

• A Base44 account (sign up at base44.com)
• An MCP-compatible client such as Claude Desktop, Cursor, or ChatGPT

Configuration

Add the following to your MCP client's configuration file:

{
  "mcpServers": {
    "base44": {
      "type": "http",
      "url": "https://app.base44.com/mcp"
    }
  }
}

Authentication

The Base44 MCP server uses OAuth. On first use, your MCP client will redirect you through the standard Base44 sign-in flow. You only need to complete this once; the token is reused for subsequent sessions.

Notes

• After adding the configuration, restart your MCP client so it picks up the new server.
• App builds run asynchronously. Use get_app_status to poll for completion and get_app_preview_url once a build is ready.
• All actions that modify or create projects require explicit user confirmation through the assistant.

• Prototype an internal tool (CRM, task tracker, inventory system) by describing it to your AI assistant and letting it call create_base44_app.
• Iterate on an existing Base44 project from your IDE: ask the assistant to add fields, pages, or workflows via edit_base44_app.
• Audit your account by listing all apps with list_user_apps and inspecting their data models with list_entity_schemas.
• Pull live data out of a Base44 app for analysis or reporting using query_entities without leaving your chat or editor.
• Automate a build-and-share workflow: trigger app creation, poll get_app_status, then surface the get_app_preview_url link to teammates.

• "Create a new Base44 app for tracking customer onboarding tasks with stages, assignees, and due dates."
• "List my Base44 apps that have 'CRM' in the name."
• "Show me the entity schemas for my 'Sales Pipeline' app."
• "Query the Contacts entity in my CRM app and return everyone added in the last 30 days."
• "Add a 'priority' field to the Tasks entity in my onboarding app and then give me the preview URL."

• Officially hosted and maintained by Base44, so it stays current with platform features.
• OAuth-based authentication means no manual API key management.
• Covers the full lifecycle: create, edit, list, inspect schemas, and query data.
• Works out of the box with any HTTP MCP client (Claude, Cursor, ChatGPT, etc.).

• Requires a Base44 account; functionality is scoped to apps inside that platform.
• Builds are asynchronous, so workflows must poll get_app_status rather than getting immediate results.
• Documentation is concise and lower-level details (full parameter schemas, error semantics) are limited.

• Retool MCP style integrations for building internal tools from AI assistants.
• Bubble and other no-code app builders, some of which expose APIs that can be wrapped as MCP servers.
• Community Base44 servers such as maorfr/base44-mcp for self-hosted experimentation.