arkanjo-tool/arkanjo ? reverse-engineered prompt

Build me a C++ command line tool that scans a codebase and finds duplicated functions.

I want it to let me explore duplicate code, search for functions by name or substring, and generate a report that shows duplication counts broken down by folder. It should work with a similarity threshold from 0 to 100, so I can decide how strict the matching is, and it should behave like near exact matching at the high end and more relaxed matching at the low end. Please make it use a preprocessing step and cache the results so repeated queries are fast, and have it automatically run preprocessing if I forget. It only needs to support languages that Tree sitter can parse, and I’m fine with those being listed in a config file.

Please include a simple build and install flow, a few clear CLI commands, and basic tests so I can verify it works. If you need to check current docs for Tree sitter or CMake details, go ahead and look them up online.