Lovable
@Lovable
Today we’re releasing the Lovable MCP server in Research Preview. Create, iterate on, and deploy apps directly from your terminal or AI agent. Available now. https://t.co/GdNb4bwt8R
32retweets344likes
View on XLovable Launches MCP Server to Build Full Stack Apps From Any Agent
· Updated
Lovable, an AI app builder that generates full-stack web applications from natural language descriptions, launched its official MCP server in research preview. This allows external AI agents like Claude Code and Cursor to manage the entire development lifecycle without opening the Lovable web interface.
- Availability
- Research Preview, Pro and Business plans
- Supported clients
- Claude Desktop, Cursor, Claude Code, and more
- Core tools
- create_project, send_message, deploy_project
- Authentication
- OAuth or API key
- Database access
- PostgreSQL (Lovable Cloud) via SQL queries
The update shifts Lovable into a programmable infrastructure layer. While the platform previously introduced Lovable's desktop app MCP support, this release makes Lovable's own engine a discoverable tool. This allows developers to combine Lovable's high-speed UI generation with the specialized reasoning of their preferred coding assistants.
You can now use terminal prompts to trigger project creation, iterate on features, and publish live apps. The server provides tools for code inspection, following a pattern seen in Lovable's Wiz security integration for automated vulnerability scanning. The research preview is available now for Pro or Business plans.
Still wondering? A few quick answers below.
The Lovable MCP server is a standardized interface that uses the Model Context Protocol to connect external AI agents to the Lovable platform. It allows tools like Claude Desktop or Cursor to programmatically create, edit, and deploy full-stack web applications. Users can manage their entire development workflow through natural language commands without leaving their preferred coding environment.
Chat connectors allow the internal Lovable agent to reach out and connect to your external tools, such as Notion or GitHub, during a build. In contrast, the Lovable MCP server works in reverse by allowing your external AI agents to connect to and control Lovable. This enables external assistants to drive the app-building process programmatically from outside the Lovable UI.
The Lovable MCP server is currently in research preview and available to users on Pro or Business plans. It is not available to the Free tier during this initial testing period. Eligible users can authenticate using either a standard browser-based OAuth flow or a workspace-specific API key for headless and scripted use cases.
The server is compatible with any AI client that supports the Model Context Protocol. This includes popular developer tools like Cursor, Claude Desktop, and Claude Code. Using the streamable HTTP transport standard, any custom-built agent or automation platform, such as n8n or LangGraph, can also connect to manage Lovable projects and deployments through structured tool calls.
Yes, standard Lovable credits apply when using the MCP server to perform specific development tasks. While administrative actions like listing projects or checking workspace status are free, any tool calls that trigger the Lovable agent to build or iterate on code, such as creating a project or sending a message, will consume credits from your workspace's existing build credit pool.
