MCP Server

Time series forecasting inside your MCP client

The TSFM.ai MCP server exposes model listing, forecast execution, and scheduling as MCP tools. Use it from Claude, Cursor, or any MCP-compatible client.

Claude compatibleCursor compatibleAny MCP client
MCP configclaude_desktop_config.json
{
  "mcpServers": {
    "tsfm": {
      "command": "npx",
      "args": ["-y", "@tsfm/mcp-server"],
      "env": {
        "TSFM_API_KEY": "your-api-key"
      }
    }
  }
}

One config block gives your MCP client access to all forecasting tools.

Tools

Forecast, list models, create schedules — all as MCP tools

Clients

Works with Claude Desktop, Cursor, and custom MCP hosts

Setup

One config block and an API key

What the MCP server provides

Each tool maps to an API endpoint. Your MCP client discovers them automatically.

Forecast tool

Run point forecasts and prediction intervals on any series. Specify the model, horizon, and quantiles directly from your conversation.

Model listing tool

Browse available models with metadata — context length, latency, pricing — so you can pick the right model for each task.

Schedule tools

Create, list, and manage recurring forecast schedules. Automate forecasts that run on a cron without leaving your MCP client.

View schedule docs

Get started in three steps

  1. 1

    Get an API key

    Sign up for a free account and generate an API key from your account dashboard.

  2. 2

    Add the MCP server config

    Add the TSFM server block to your MCP client config file. The server handles tool registration automatically.

  3. 3

    Start forecasting

    Ask your MCP client to forecast a time series. It will discover and call the TSFM tools with no additional setup.

Frequently Asked Questions

Add forecasting to your MCP client

One config block and an API key. Start forecasting from Claude, Cursor, or your own MCP host.