LiamPrevelige

For me it’s usually some context-specific, low-level DAG of functions/data flow. Maybe a higher level of C4 model depending on the scope of my work.

I’ve been studying this to build a VS code extension that visualizes my codebase as I work. Similar to Feynman’s point, there are many ways of visualizing the same task/section of code, but I’ve found some are more universally interpretable than others.

There seems to be a bug or comment depth limit, I can't reply to your comment below. I just published an update to extension description explaining that code moves through our servers. It may take a few minutes to appear. Since this post, we added a privacy policy to our website and the ability to input your own API key, circumventing our servers altogether. I will list this feature in the extension description once we have time to require manual triggering of architecture generation

Agreed this is a poor experience, we're making pricing very transparent now. Hope you give CodeViz another chance. Were you able to generate the top level architecture or interact with the call graph? If you couldn't use CodeViz, or saw the subscription modal immediately after opening the extension, this is a bug that I'll look into

Thanks for the report, following up now. We'll make some updates to our website asap. Preview of the call graph can be seen in our video demo: https://www.youtube.com/watch?v=fgfDXUtWzRk. Happy to chat about reduced subscription price.

I created an issue for user provided API keys, tracked here: https://github.com/EdisonLabs-Inc/CodeViz-Public/issues/3

To generate second layer of architecture diagram, $19/mo subscription with a 7 day free trial. The call graph and top layer of the architecture diagram are free. We'll update our website and the extension with pricing info asap. The subscription didn't exist until a couple of days ago, since layered architecture is a feature that we made public specifically for HN to try.

We only use Anthropic at the moment, but we left OpenAI in the Data & Privacy in case we change models (not in the foreseeable future)

> The only way I found out was installing it, create a call graph, and the first click took me to a payment page

Yes, we need to be much more clear. It's not intentionally hard to find, we just been distracted with bugs and haven't had time to update the website. The subscription didn't exist until a couple of days ago. The layered architecture is a feature that we made public specifically for the HN community

> Having to fill out my payment details for a trial version is a a hard no for me

We'd love to embed a demo repo for people to play with, or let users generate nested architecture for one repo before having to subscribe

Thanks for providing context on the LLM approval process! User provided API keys will be completed by the end of the week: https://github.com/EdisonLabs-Inc/CodeViz-Public/issues/3

> The price is too steep. I get way more value out of Copilot than I get from CodeViz which is similarly priced.

Copilot is a great product and inexpensive, I do understand where you're coming from. Personally, I spend most of my time reading code and gathering context vs typing code. We hope to provide much more than $19/mo in value, so the subscription is a good signal for whether we're making something useful for you. We also have more features for team licenses that we haven't publicized.

The reduced subscription price for using your own API key does make sense though. If you send me an email or message me on discord I'd be happy to chat this over. This goes for anyone with a strong opinion on the matter

Should we let users choose which folders they want analyzed and leave all selected by default?

On it! Tracked here: https://github.com/EdisonLabs-Inc/CodeViz-Public/issues/2

Appreciate the honest feedback! We're working hard to improve CodeViz, please let me know if there are any features you'd love to see.

I cloned pulumi and exported both layers of the CodeViz diagrams into mermaid format: https://github.com/EdisonLabs-Inc/Pulumi-Diagrams/tree/main. I'm not familiar with the repo so let me know if anything looks off. Hope you find these useful!

Glad you like it! Let us know if you have any suggestions

We'd love to have CodeViz run with local LLMs so you can untangle the spaghetti. The call graph is generated without LLMs or servers - any interest in a 'local only' mode so you can still use CodeViz while keeping your code on-device?

Sure thing - we'll plainly state this in the extension's description, the extension itself, and our website

I get where you're coming from - Copilot is very cheap and useful! We want the free version of CodeViz to be as useful as possible for you. Some features, like the second architecture layer, are just expensive for us.

Personally, I spend a small fraction of my time actually typing code and much more time gathering context & building a mental map. CodeViz speeds up the mental map part, so we hope to deliver much more value than $19/mo

Let me know if there's anything missing from CodeViz that would change your mind!

Got it, how about a manual trigger for diagram creation and an explicit description of LLM/server usage on the page with the trigger?

Some text on our website's homepage (https://codeviz.ai)

We definitely want to standardize diagram representation, especially with deterministic representations of code. I really like the visual notations in UML diagrams, but in my opinion they often contain too much text. What do you think? In general, we want to put the same density of information that a dev has in their head into the extension.

Did you get a chance to try the call graph? It's a version of the function call mapping, although the layout and UX needs some improvement. Planning on: 1. Having all functions appear on the graph. As you click around in VS Code, the camera pans to the section of the graph you're working on 2. Let the user choose their scope (functions, classes, files, etc) 3. Unify architecture view with call graph, with a minimap to quickly jump between layers

Would love to hear your thoughts (and anyone else's) on this

Also, I really like sequence diagrams, though they're a bit tough with the dimensions of the extension. Maybe we can contain similar info in the function/class mapping, by using a hierarchical layout & edge labels

Just added an option to choose between Google and Apple calendar

The name association is a genuine concern of ours :)

That's correct, our server just routes calls directly to Anthropic. Some users requested an option to input their own API key and talk to Anthropic directly. I'll add this by the end of the week, maybe today if time.

Local LLM is still the end goal

Thanks! Inspired by my bad memory and half of our website visitors being on mobile

Yes, some code is sent to Anthropic. We're hoping to: find initial users that are comfortable with software copilots (github copilot, cursor, etc) -> iterate on their feedback to make diagram generation require less sophisticated LLMs -> move everything locally

We'd love to use local models and have played around with them a bit. Exactly right about the labeling - we didn't stick with local models because 3.5 sonnet is exceptionally good at finding niche architecture labels and merging similar modules (since code analysis is chunked). Copilot tools are becoming very popular, so companies are getting less strict around LLMs and code, but ultimately we do think everyone is better off if CodeViz is self-contained.

There is some hard coded bias for web dev. Diagram modification is definitely high on our todo, and we've been finding ways to reduce pre-defined structure in our prompts to LLMs so they work with broader tech stacks. When we sell licenses to teams, we do some manual checks for accuracy and detail, which helps us improve the public extension.

What's the name of the robotics repo? And any preference for modifying the diagram directly vs instructing changes by text?