timesfm-mcp

Give any AI agent time-series forecasting superpowers.

timesfm-mcp is an MCP server that lets Claude Code, Claude Desktop, Cursor, or any MCP-compatible agent forecast a series of numbers — sales, traffic, usage, costs — and reason about the result.

Why

LLM agents can read, write, and run code — but they can't see the future. This gives them a clean forecast tool. The agent calls it, gets point forecasts + uncertainty bands + a compact trend/seasonality summary, and writes the explanation and recommendation itself.

What you get

Tool	What it does
forecast	Forecast a single series with optional uncertainty bands
list_backends	Report which engine is active (timesfm / baseline)
backtest	Hold out the last N points and compare TimesFM vs baseline (MAE/sMAPE)

Two backends, zero configuration

Backend	When active	What it needs
Statistical baseline	Always	Just numpy — already a dependency
TimesFM 2.5 (Google)	When installed	pip install "timesfm-mcp[timesfm]"

The server auto-detects TimesFM and uses it; otherwise it falls back to the baseline. Either way, forecast always returns something useful.

Quickstart (30 seconds)

uvx timesfm-mcp

Add to your agent config:

{
  "mcpServers": {
    "forecast": { "command": "uvx", "args": ["timesfm-mcp"] }
  }
}

Then ask your agent: "Forecast the next 6 months from this revenue data and tell me what to expect."

Get started → See client configs →