Configuration guides for IDEs, desktop clients, and cloud services like Flowise.
The Chatmu MCP Server natively supports two parallel connection methods depending on your AI client's capabilities: HTTP/SSE (Server-Sent Events) for modern Cloud IDEs and autonomous agents, and Stdio (Standard Input/Output) for strictly local desktop applications.
This is the recommended, universal standard for most modern applications. Because the Chatmu MCP Server is hosted securely in the cloud, you do not need to install anything locally. You simply provide the public Server URL along with your API Key.
Navigate to your Cursor Settings (Features -> MCP) and add a new MCP Server using the following JSON configuration:
{
"mcpServers": {
"chatmu-cloud": {
"serverUrl": "https://mcp-chatmu.aaatmi.com/mcp?key=YOUR_CHATMU_API_KEY"
}
}
}
https://mcp-chatmu.aaatmi.com/mcp?key=YOUR_CHATMU_API_KEYNote: If your client natively supports injecting headers, you can also omit the ?key= parameter and instead pass the key dynamically via the Authorization: Bearer YOUR_CHATMU_API_KEY HTTP header.
If your AI ecosystem is strictly limited to local execution and does not currently feature a URL input field (such as the official Anthropic Claude Desktop app), we provide a universal NPM bridge. This bridge executes locally but seamlessly translates the Stdio commands directly to your cloud instance.
You will need to modify your client's claude_desktop_config.json (or equivalent configuration file) to execute our global NPM package using npx:
{
"mcpServers": {
"chatmu-cloud": {
"command": "npx",
"args": [
"-y",
"@aaatmi/chatmu-mcp"
],
"env": {
"CHATMU_API_KEY": "YOUR_CHATMU_API_KEY"
}
}
}
}
How it works: The npx -y command automatically fetches the latest lightweight bridge without asking for installation prompts. It will securely read the CHATMU_API_KEY from the environment context and establish a permanent tunnel to your remote capabilities.