I’m exploring ways to measure productivity gains from AI-assisted coding and found this repository intriguing. My proposal is to add functionality that differentiates between AI-generated code and code manually written by developers. For instance, since Roo Code applies diffs directly in VS Code, it should be feasible to detect and label code segments created with AI tools. This would enable metrics such as the percentage of AI-generated code relative to total code produced.
Has there been any discussion on this feature? Additionally, do you think it could be integrated within the existing ecosystem?