I hope it has better quality than ManusAI

I would genuinely start with assessing which tech debt exists, this does not require LLM (e.g. use Enola). Then think, whether it's worth fixing.

Allowing an LLM to solve the tech debt will most likely lead to micro-optimisations, and in larger repos even very odd architectural patterns.

Very interesting. Will check this out!

Have a look at Enola and share us the feedback. We know that building a performant deterministic model of codebase is a big challenge. While there are AST tree-sitters, they are simply not enough. A lot more work needs to go in.

When you publish your project, ping us!

Uh.. I mean can someone tell me these days when it's not written by AI.