logoalt Hacker News

Show HN: Nogic – VS Code extension that visualizes your codebase as a graph

50 pointsby davelradindratoday at 6:43 PM19 commentsview on HN

I built Nogic, a VSCode extension currently, because AI tools make code grow faster than developers can build a mental model by jumping between files. Exploring structure visually has been helping me onboard to unfamiliar codebases faster.

It’s early and rough, but usable. Would love feedback on whether this is useful and what relationships are most valuable to visualize.


Comments

kachapopopowtoday at 11:16 PM

I always liked the idea having relationship based programming (graph programming), but with actual code. Never actually made the effort to make something like that. Pretty neat either way.

pentaphobetoday at 9:43 PM

All the GitHub links on your extension page are borked (including issues)

From the look of the associated domain it looks like you're going full product, best of luck

I'm a huge proponent of graph & visual analysis of complex systems - would have loved to try this out, but will always skip closed source editor extensions (especially in the age of widespread npm supply chain attacks & vibe coding)

tiborsaastoday at 10:59 PM

I've tried it, but it's very slow on a not too complex codebase with my M3 Macbook Air.

    --------------------------------------------------------------------------------
     Language             Files        Lines        Blank      Comment         Code
    --------------------------------------------------------------------------------
     Typescript JSX         121        18724         1699         1051        15974
     TypeScript              61         5389          629          550         4210
     CSS                      5         1039           50           22          967
     Markdown                 3          657          173            0          484
    --------------------------------------------------------------------------------
It was like 5-10FPS at best, not really usable unfortunately because I like these tools.

I'm using another similar one which is buttery smooth, Code Canvas.

vmware508today at 10:23 PM

I guess it is still useless in Ruby or Ruby on Rails. Standard "find the method declaration" or "used here" do not work in Ruby on Rails. Still, huge companies maintain that Ruby on Rails mess, where you cannot properly investigate, so you just guess and use the search and find option. Those codebases won't be replaced for a while, but good luck working on them. Such a headache!

show 1 reply
Scene_Cast2today at 11:07 PM

I used to use Doxygen to create caller and callee graphs to understand code flow. Unfortunately, the tool hasn't really changed in more than a decade.

patabytetoday at 10:20 PM

I've been having great success with LLMs generating Mermaid diagrams and flowcharts from a repo. Claude Code and Cursor both do consistently great jobs. For example: `generate a mermaid swimlanes diagram of the XX logic flow`.

nebula8804today at 10:03 PM

Very cool visualization. However it crashes on a more complex project. I added a folder with 2000+ files(included my assets) and now the visualizer locks up then shows nothing on its tab in VSCode. How do I manually delete old boards so that I can try again with a smaller slice of the code(without assets)?

everliertoday at 9:22 PM

Nice, I wanted to build something similar for a long time. The coolest thing is to start summarising clusters for very large codebases, which essentially provides an LoD system for the context.

Imustaskforhelptoday at 10:03 PM

Funny how world is so tiny. I am literally building myself an vscode extension which can abstract an api on top of google colab's vscode extension and I am able to effectively create a sandbox for any python code (I mean to be fair they all still share the same resource but that resource is of google)

I have also hacked together a way for it to create new kernels aka new vm's itself but that becomes really really slow and also I am trying to look at other options to sandbox inside the jupyter notebook itself.

The end result was very messy though so I was literally just currently experimenting with if I could just scrape/automate it from the browser directly.

All in All I must admit that Vscode extensions are/feel very quite competent from what I can gather.

puppycodestoday at 10:53 PM

I definitly think more tools like this are needed, but not open sourcing it is a mistake.

You will be quickly replaced by a friendlier competitor.

fpausertoday at 10:00 PM

Closed source vscode extensions: not for me.

suprjamitoday at 10:38 PM

Only JS, TypeScript, and Python. You got me all excited for a C visualizer!

fishgoesblubtoday at 10:02 PM

Unfortunately the repository links are broken and this is ARR licensed.

wektoday at 9:49 PM

From your first page, this looks cool and needed. But as others have posted, I can't get to your github pages.

Aspostoday at 9:46 PM

The illustration gif is way too fast. Hard to understand what is going on. Slow it down 2x or so.