bbak/mcs-mcp

Implements an MCP server that reconstructs Jira workflow history as event-sourced data, then applies statistical techniques—Monte-Carlo simulation, XmR control charts, and Little's Law analysis—to forecast delivery timelines and detect process bottlenecks stratified by work type. Automatically infers workflow semantics (demand funnel, active work, terminal states) and supports backtesting forecasts against historical replays to validate predictive accuracy. Prioritizes data minimization by ingesting only issue keys, status transitions, and timestamps while dropping titles and descriptions to prevent sensitive content leakage to AI agents.