Built an AI tool to visualize large codebases - would love feedback
I’m a solo technical founder and just launched the first public version of Codemap AI (https://code-map.com) - a tool to help developers instantly understand large codebases.
Codemap AI currently:
- Parses large codebases (Java for now)
- Generates interactive relationship diagrams
- Lets you chat with an AI about your code
- Lets you chat with dev agents (developer clones) about code changes
- Tracks security issues and project statistics
- Helps with onboarding, debugging, and documentation
I built this because I’ve personally spent weeks trying to understand unfamiliar and legacy code in large projects. My goal is to make codebases as explorable as possible - like having a "Google Maps" for your code.
At this stage, it’s not a live SaaS - I have a landing page and a demo video here:
- Landing page: https://code-map.com
- Demo video: https://youtu.be/kMy-zwsApxI
Would love feedback from the HN community:
- Does this solve a real pain for you?
- What features would make it indispensable?
- Any suggestions for improvements or integrations?
Thanks - happy to answer any questions!
And think about renaming the "Bugger Agent" ... Bugger has a particular meaning in English that you won't like.
And we’re starting to onboard small teams for testing - happy to chat if this could be useful for your team!
Thanks for the feedback!
does code-map requires license from major model provider?
also, is code-map open source? how are new languages added?
Good questions — here's some clarification:
1. License / model dependency Codemap doesn't require a license from a major model provider - it's designed to work with local models (Ollama, open-source LLMs), or users can plug in external APIs (OpenAI, Claude) if they want. The core parsing/analysis/graphing is independent of any one model.
2. Open source? Codemap itself is not open source - it's a commercial B2B product. Some parts (like parsers) may use OSS components, but the overall system is proprietary.
3. How are new languages added? Language support is modular: each new language needs a parser + chunker to feed the AI layer. Java is fully supported today. JS/TS are next. Python, PHP planned.