mcp-remote bridge

For any client that only speaks stdio.

Kind Bridge

Transport A local stdio process that proxies to the HTTP endpoint.

Config —

Setup

Use the block below anywhere a client takes a stdio command.
Restart the client.
If the tools do not appear, run the same command in a terminal — its errors go to stderr, which the client usually hides.

{
  "mcpServers": {
    "unzoi": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote", "https://api.unzoi.com/mcp",
        "--header", "x-api-key:${UNZOI_KEY}"
      ],
      "env": { "UNZOI_KEY": "nai_your_key_here" }
    }
  }
}

What differs here

This is a workaround, not an upgrade. If your client speaks remote HTTP, use it directly — the bridge adds a process, a Node dependency and a place for the key to leak into a log.
It is the standard fallback for desktop clients on plans without custom connectors, and for editors whose MCP support predates Streamable HTTP.
Running the same command by hand is the fastest way to debug it, because a client that spawns it usually swallows stderr.

Next

Get a key at console.unzoi.com, if you have not.
The tool reference — what the model is reading when it chooses between the six tools.
Stories for context windows — the one habit that most changes how well an agent works against this API.
Limits — an agent that loops can spend a monthly quota faster than you expect.
mcp-remote bridge's own MCP documentation — authoritative for anything about the client itself, including config paths, which move.

Endpoint, for copying: https://api.unzoi.com/mcp