Show HN: Komment – make sense of messy codebases (komment.ai)

I was an architect in a big tech company for years and worked with a lot of different teams, worked on a lot of different codebases all with varying degrees of documentation and complexity. What me and the team at Komment have tried to build is a way to make sense of projects, whether it's something we've built or others or something we've to get our heads around.

It:

- connects to a Git repo

- processes the codebase with a combination of deterministic exploration, static analysis and dependency graph generation

- Uses a few different LLMs (depending on language, purpose, etc.) to generate a structured wiki with architecture diagrams, usage examples and external dependencies.

Our current thinking is that it would be useful for agencies who have to produce documentation for clients or for clients who receive delivery of external code but if you have any other ideas, I would love to hear them!

Open to feedback, skepticism, and ideas on making this better.