I've been using this for a while, and I thought others on HN might find it helpful. You can run it against a local repository, and it uses mostly non-AI tools to make a high-level assessment about how AI-native the repository is. This is all based on my own personal experience and is very much opinionated, but this has helped me grow my codebase without it collapsing under the weight of the AI coding biases that tend to happen unless you put the guard rails in place. This is me trying to structure what I've learned into something that can be easily applied to any codebase relatively quickly. It also comes up with a nice vector graph image, which I think is quite cool.