I was inspired and wanted to play around with the idea, both for learning and also to try it on my code, so I built my own implementation of it (I could not find the sources from the OP).
It was important for me to make it as a local and privacy respecting app (i.e. running locally), which placed some constraints on the model size I could use, but the results are looking promising.
Feel free to play around with the tool / code (this is why I'm sharing it)! I would love to hear if it detects anything interesting for you.
1 comment
[ 2.9 ms ] story [ 14.9 ms ] threadI was inspired and wanted to play around with the idea, both for learning and also to try it on my code, so I built my own implementation of it (I could not find the sources from the OP).
It was important for me to make it as a local and privacy respecting app (i.e. running locally), which placed some constraints on the model size I could use, but the results are looking promising.
Feel free to play around with the tool / code (this is why I'm sharing it)! I would love to hear if it detects anything interesting for you.
Here's a demo output: https://sturdy-dev.github.io/suspicious/demos/cli_py/ (more in the README.md)