Agentblame: Line-level AI attribution using Git notes (github.com)

A new tool called Agent Blame has been introduced to help developers better attribute AI-generated code within their GitHub repositories. This tool facilitates the identification of lines of code written by AI by marking them with orange indicators in Pull Requests (PRs) and leveraging a command-line interface (CLI) that shows AI contributions in any file. By integrating seamlessly with popular AI coding assistants like Cursor and Claude Code, Agent Blame captures attributions automatically, preserving this information even through complex Git operations such as squash and rebase merges. This development is significant for the AI/ML community as it enhances code review processes by clarifying how much of the codebase is generated by AI versus human efforts. Such transparency could lead to more informed discussions about code quality, ownership, and the ethical implications of using AI in software development. Key features include real-time attribution tracking, a user-friendly Chrome extension for easy access, and GitHub Actions that ensure attributions are maintained during merges. The tool is built using Bun and Git 2.25+, and it encourages community contributions for future expansions, like support for other coding agents and additional functionalities.