Hey! Love the project. I've been using it so much 🙏
One thing I keep wanting is a way to actually see the graph. Like, I know the nodes and edges are there in SQLite, but there's no way to explore them visually. Would love to click around and understand how my codebase is connected and see a visual representation.
What I'm thinking
A new codegraph studio command that spins up a tiny local server and opens a browser-based graph view. Click a node, see its callers/callees and a source snippet. Search to jump to a symbol. Expand neighbors on click instead of rendering everything at once (important for large codebases).
Stack-wise I was thinking:
- Single HTML file, no bundler
- Cytoscape.js for the graph (handles 10k+ nodes, lightweight)
- Small Express/http server reading directly from the SQLite db
- Color-coded by symbol type (function, class, method)
The feature I think would make it genuinely useful over a generic visualizer: since CodeGraph already has impact radius + caller/callee traversal, the studio could highlight "what breaks if I change this?" as a subgraph. That's actually actionable.
Before I build it
Just wanted to check: does this fit the direction you have in mind for the project? Any preferences on the command name, approach, or things to avoid? Happy to open a PR once we're aligned.
Thanks!
Hey! Love the project. I've been using it so much 🙏
One thing I keep wanting is a way to actually see the graph. Like, I know the nodes and edges are there in SQLite, but there's no way to explore them visually. Would love to click around and understand how my codebase is connected and see a visual representation.
What I'm thinking
A new
codegraph studiocommand that spins up a tiny local server and opens a browser-based graph view. Click a node, see its callers/callees and a source snippet. Search to jump to a symbol. Expand neighbors on click instead of rendering everything at once (important for large codebases).Stack-wise I was thinking:
The feature I think would make it genuinely useful over a generic visualizer: since CodeGraph already has impact radius + caller/callee traversal, the studio could highlight "what breaks if I change this?" as a subgraph. That's actually actionable.
Before I build it
Just wanted to check: does this fit the direction you have in mind for the project? Any preferences on the command name, approach, or things to avoid? Happy to open a PR once we're aligned.
Thanks!