hey HN!<p>There's lots of AI Code Review tools by now. Our experience with these tools has been quite poor so far so we decided to tackle this space differently.<p>We started building Callstack as a deep code understanding engine (you can check out the UI here: <a href="https://graph.callstack.ai" rel="nofollow">https://graph.callstack.ai</a>) to understand complex codebases to a granular structural level. The premise for this effort is to serve an LLM with a deterministic context retrieval over a complex codebases which impacts the performance by a large margin.<p>With few experiments this enabled us to perform really well over simple tasks so we have shipped code reviews, talked to users and today it runs in production at companies of all sizes.<p>Today we are launching self-service for the Code Reviewer. It's completely free for individual use and open-source repos. It's also fully private, it runs as a Github / Gitlab action (the only 3rd party API request is to Anthropic) and supports 9 major languages.<p>You can try it out in less than 3 minutes with no credit card required, just install it on your Github / Gitlab repo.<p>Would love to know your feedback!