io.github.doazvjettu/leakguard-mcp

Analyze this quantitative backtest code and check for lookahead bias, future data usage, train-test leakage, and feature engineering issues that could distort results. Report findings with locations, risk explanations, and fixes.

A static analysis report identifying suspicious code locations, leakage types, impact explanations, and recommended fixes.

Inspect this machine learning feature engineering and data split code for leakage introduced by scaling, imputation, label construction, or temporal splitting, and explain the correct processing order.

A checklist of leakage risks plus rewrite suggestions that preserve temporal order and proper training flow.

Before running the backtest, perform a static check on this strategy code and flag any logic that may access future prices, future labels, or full-dataset statistics, ranked by severity.

A severity-ranked issue list that helps fix high-risk flaws before backtesting.