I just open-sourced a lightweight ML-based bot detection API. It uses simple features like user-agent entropy, path entropy, outdated browser patterns and sensitive URLs to detect automated requests.

Built with LightGBM and Flask No IP tracking, no cookies Ideal for edge use cases or form validation Easily extensible via feature hooks Model can be retrained from logs (retrain_model.py)

https://github.com/TimTom32hh/ml-bot-score-api

Would love feedback or contributors interested in privacy-first bot defense.