How to use MCP in Retool: Two setup options for AI agents

Enterprise AppGen introduces AI-powered app generation that is fast, secure, and production-ready, leveraging the Model Context Protocol (MCP) introduced by Anthropic in 2024, which has become a popular method for providing context and data to AI agents from various systems. MCP servers enable large language models (LLMs) to access and interact with tools like GitHub, Figma, and Jira, enhancing their functionality and utility. While some services like Atlassian have public MCP servers allowing easy integration with AI tools, many MCP servers are available only as GitHub repositories, requiring users to host them independently. Platforms like Smithery.ai facilitate the deployment of these servers by allowing users to configure necessary parameters through smithery.yaml files or URL parameters, enabling integration with tools like Retool to create powerful AI agents. With a public URL, an MCP server can be set up as a resource in Retool, allowing AI agents to access and use the associated tools, significantly expanding their capabilities.