Improve feedback loop for AI agents with custom linter

This article discusses the importance of improving feedback loops for AI agents in software development. It highlights that static instructions often fail to prevent coding violations, such as inline imports. The author advocates for automating quality checks with custom linting tools that AI agents can execute, providing immediate feedback. The article introduces jg-lint, an extensible Python linter with a Rust core, and demonstrates how to implement custom rules using AST parsers to enforce team-specific coding standards. This approach aims to reduce manual code review and enhance AI agent output quality.