AI clients
Bring your data to the assistant you already use
The MCP server speaks the standard protocol over stdio, so it works with Claude Desktop, Claude Code, Cursor, Windsurf, and other MCP clients.
Setup by client
Pick your client. Every setup takes a few minutes and one MCP token.
Claude Desktop
The most common setup. One config entry, then restart the app.
- 1In the dashboard, open Data sources and issue an MCP token. Copy it — it is shown once.
- 2In Claude Desktop, open Settings → Developer → Edit Config to open claude_desktop_config.json.
- 3Paste the configuration below, replacing the placeholders with the values from the dashboard.
- 4Quit and reopen Claude Desktop. The tools appear under the tools menu in a new chat.
{
"mcpServers": {
"growithamazon": {
"command": "npx",
"args": ["-y", "https://<your-api>/mcp/client.tgz"],
"env": {
"GROWITHAMAZON_API_URL": "https://<your-api>",
"GROWITHAMAZON_MCP_TOKEN": "<token from the dashboard>"
}
}
}
}Keep the server name and the GROWITHAMAZON_* variable names exactly as shown — the MCP client reads them. The dashboard fills in your real API URL when you issue a token.
Good to know
One token per client
Issue a separate MCP token for each person or app, so you can see usage and revoke one without affecting the others.
Same workspace, every client
All clients see the accounts connected to your workspace, including connections teammates share.
Nothing to maintain
Token refresh and account selection happen on the server. The client only needs its MCP token.
Approve calls as you go
Most MCP clients ask before running a tool. Approve reads once; review write actions every time.
Ask your first question in ten minutes
Create a workspace, connect one account, and add the MCP server to your assistant. No API keys to paste and nothing to export.