Satvik-jain/trainlint: Offline checker and agent skills for silent ML bugs

Trainlint is a pure Python offline checker and collection of agent skills designed to catch common, silent machine learning bugs that do not crash execution, such as data leakage, incorrect training modes, and missing random seeds. The tool parses code using the ast module without running it or touching a network or GPU. It provides fourteen rules and four agent skills to help developers and coding assistants avoid these silent errors. Open article →