Some AI tools don’t allow the use of remote MCP servers. While Flowglad only
has a remote MCP server, we detail a simple workaround below.
You’ll need a Flowglad API key to authenticate to the MCP server which you can
get from settings.
Remote Server Setup
The Flowglad MCP server can be found at the following URL:
Cursor
Cursor
Open Cursor’s
mcp.json file through the command palette with Cmd/Ctrl + Shift + P or through Cursor Settings > Tools & MCP and paste in the following under mcpServers. Make sure to replace <FLOWGLAD_API_KEY> with your Flowglad API key.mcp.json

Claude Code
Claude Code
Run the following command (make sure to replace You can check that this worked by first running
<FLOWGLAD_API_KEY> with your API key):claude, to open Claude Code, and then use the /mcp command to see a list of MCP servers that Claude Code has access to, which should now include Flowglad’s.Local Server Workaround
This workaround uses mcp-proxy to allow tools that prohibit the use of remote MCP servers, like Claude Desktop, to utilize Flowglad’s remote MCP server.1. Install mcp-proxy
First, you’ll need to install either uv or pipx which are package managers for Python if you don’t have them installed already. After you have one installed, you can find themcp-proxy installation documentation here.
After you’ve installed mcp-proxy, run which mcp-proxy to print the path of the mcp-proxy binary and copy it. We’ll need this path in the following steps.
2. Add MCP Config to Your Tool
Here’s an example of a Claude Desktop config using this workaround. You can edit this config under Settings > DeveloperMake sure to replace
<MCP_PROXY_PATH> with the path you got from running which mcp-proxy above and <FLOWGLAD_API_KEY> with your Flowglad API key.