Skforecast AI
Automate time series forecasting pipelines with deterministic modeling decisions, standalone reproducible Python code, and an MCP server for coding agents.
Skforecast AI is a transparent time series forecasting assistant built on the skforecast library. It automatically profiles time series data, selects optimal forecasters and estimators using deterministic, rule-based algorithms, and provides prediction intervals alongside robust backtesting evaluations. Instead of acting as an unpredictable black box, it guarantees full reproducibility by outputting the exact, standalone skforecast Python code that produced the forecast. An optional LLM integration powered by Pydantic AI explains decisions and translates natural language instructions into backtesting schemes without directly making the modeling choices. With built-in CLI functionality and a Model Context Protocol (MCP) server, it seamlessly extends AI coding agents like Claude Code and Cursor with explainable forecasting capabilities without exposing sensitive datasets.