Agent-ML-skills – Teach Codex/Claude/Cursor to stop making ML mistakes (github.com)

A new initiative called Agent-ML-skills has been launched to enhance the capabilities of AI coding agents like Codex, Claude, Cursor, and OpenCode in the field of Machine Learning (ML) and Data Science. This curated set of 15 practical skills addresses common pitfalls in ML development, such as preventing data leakage during preprocessing and ensuring appropriate evaluation metrics for imbalanced datasets. The installation of these skills can be done effortlessly through a single command, streamlining the process for developers looking to improve their agents' performance. This tool is significant for the AI/ML community as it empowers coding agents to adopt best practices typically followed by seasoned ML engineers, thereby reducing errors and boosting productivity. The skills also promote reproducibility, efficiency, and effective model evaluation, ensuring that agents can navigate complex ML workflows seamlessly. With a focus on practical, concrete code patterns, Agent-ML-skills aims to help users avoid common mistakes and foster better collaboration between humans and AI in coding environments.