# PyOD

A Python library for anomaly and outlier detection across tabular, time-series, graph, text, image and audio data, with more than 60 detectors and a benchmark-backed orchestration engine.

PyOD is a Python library for anomaly and outlier detection across tabular, time series, graph, text, image and audio data. It provides more than 60 detectors and a benchmark-backed engine, ADEngine, for orchestrating them.

Version 3 adds an agentic workflow: an od-expert skill lets Claude Code and Codex use PyOD directly as part of an agent's own workflow, in addition to using it as a conventional Python library.

It installs as a package from PyPI or conda-forge, with documentation, automated testing and coverage tracking, and is released under the BSD-2-Clause license.

Website: https://pyod.dev/