home / mcp / weaviate mcp server
Provides an MCP server for Cursor IDE to run and test custom tools via stdio or SSE transports.
Configuration
View docs{
"mcpServers": {
"andreashornqvist-mcp": {
"url": "http://localhost:8000/sse",
"headers": {
"DEBUG": "false",
"MCP_USER_AGENT": "Cursor-MCP",
"MCP_SERVER_HOST": "0.0.0.0",
"MCP_SERVER_PORT": "8000"
}
}
}
}You can run an MCP server to connect Cursor IDE with Model Context Protocol, enabling you to build and test custom tools that respond to prompts and tasks. This server supports both local stdio and remote SSE transports, making it flexible to run on your machine or in the cloud.
Set up your MCP server and connect Cursor IDE to it to enable interactive tool usage. You can run the server locally with stdio transport or expose it remotely via SSE, then add the MCP server in Cursor Settings under Features.
Prerequisites: you need a shell environment and the ability to install software via your preferred package manager. You will use either Docker for a quick start, or a traditional Python-based setup with the uv tool, or you can run the server directly from Cursor IDE.
HTTP MCP server configuration uses a remote SSE URL, while STDIO configuration runs a local tool via a command. The following configurations show how to run both options and how to connect them to Cursor.
Configure environment variables to control port, host, and debugging. Keep sensitive values secure and avoid exposing the server to untrusted networks in development. The following environment variables are commonly used for MCP servers.
Test the local SSE endpoint and verify that Cursor can communicate with your MCP server by connecting to the SSE URL and sending a simple prompt. Use the included test commands to confirm the server starts and responds.
A minimal MCP tool used in examples to run a simple server via stdio or SSE transports.