Hi all,<p>We launched Reaper at the end of last year (<a href="https://www.emergetools.com/blog/posts/dead-code-detection-with-reaper">https://www.emergetools.com/blog/posts/dead-code-detection-w...</a>) with the goal of helping teams discover dead code in their mobile apps.<p>Unlike typical static analysis that only finds technically unreachable code, Reaper is an SDK that monitors production data to discover code that's unused by real users (ex. stale feature flags).<p>ReaperAI takes this a step further by actually being able to open pull requests in your repo to automatically delete the dead code that it finds.<p>Here is a demo video: <a href="https://www.youtube.com/watch?v=y2FEaAmUvNw" rel="nofollow">https://www.youtube.com/watch?v=y2FEaAmUvNw</a><p>We're here to answer your questions & would love to hear any ideas or feedback you have!
The data-to-code AI inference is inherently nondeterministic. How is that an improvement over performing deterministic coverage analysis? That sounds snarky, but it’s a sincere question.
I'd change the description from "delete" to "identify" or something. I'm not sure if this is a good idea, but I imagine that the tool will just delete the lines without human intervention and that is very scary.