bjoernaagaard/xpd
Lints Python that uses pandas, NumPy, SciPy, scikit-learn, scikit-lego, MAPIE, Category Encoders, linearmodels, arch, pymoo, plotly or Apache Hamilton for slow operations and wrong results, after each edit and through MCP tools.
Write clear, typed, maintainable Python. Use whenever you write, change, refactor, debug, or explain Python code: modules, functions, classes, type hints, dataclasses, errors, comprehensions, imports, pyproject.toml, or scripts. Trigger on Python, .py files, uv, ruff, type checkers, pandas, or notebooks turned into modules, even without the words "best practices".
Review existing Python code for correctness, types, errors, module design, readability, and maintainability, and write a prioritized report where every finding cites the rule it breaks and shows the fix. Use for a review, audit, critique, grade, or check of a .py file, package, project, diff, or pull request, or when the user asks what is wrong with their Python or how to make it cleaner, even if they don't say "review".
Python library habits: run xpd to lint code for slow or wrong use of libraries such as pandas and NumPy, profile data files, and prove refactors are equivalent.