I had been wanting to play around in a project with dense user interfaces and Bloomberg terminal aesthetics and, so an investment dashboard seemed like a good fit.
It aggregates market indicators that have been known to generate contrarian buy/sell signals. CNN Fear & Greed, Bank of America SSI, AAII Investor Sentiment Survey among others.
A few technical details:
* vibe coded ~70% of it — the parts that not were either UI polish that was faster to do directly or points of the data scraping pipeline where Claude got stuck
* architecture: vanilla JS frontend + Python/Flask backend + background jobs that run an AI data extraction pipeline (Perplexity Sonar / Exa for search + GPT-5)
* runs on render.com as web service + two cron jobs that run the data update process and send daily email notifications
At some point I did try Codex for a few PRs that were never merged and I re-did instead with Claude.
Httpdiff is a really good tool and way more powerful than mine.
I've used it multiple times and wanted to try making something that focused more on the request body (which in Httpdiff are saved to temporary files) and that was also a bit more visual than a CLI.
Right. I was aiming to do it client-side only but I found out that web browsers restrict the headers you have access to through cross-origin XHR requests [1][2].
So it was a trade-off and I preferred showing all the headers at the expense of having to proxy the calls through a server.
1) the diff is really simple, line by line and text based. I'm
just using this really awesome library to do it: https://github.com/kpdecker/jsdiff The JSON is just being pretty printed to be able to do a text diff in a meaningful way.
2) Yes, I do, and thanks for the suggestions! But I just wanted to go with the minimal version first and see if people found it useful.
I didn't know about Runscope's diffing feature. Btw, I'm using a HTTPBin for the sample requests, which is a Runscope project as far as I know :)
This is my first project in React.js and it was really nice to work with.
The only part I struggled with was while building the table editor. I'm using contentEditable there and apparently React doesn't play well with it. The virtual DOM gets out of sync after user input and you have to manage it manually.
I really enjoyed making this and I'm happy to hear any feedback about ways to improve it! :)
Just so you know, there are very few situations where one should produce a graph from a set of points by spline interpolation (at least in the natural sciences). It almost never is the right choice, if you don't have a model for how the graph you are sampling looks like you shouldn't draw lines between points at all.
I browse with cookies and localStorage disabled by default (I whitelist sites I need to log into). Your site just comes up blank for me... maybe it requires localStorage for some reason? Not sure why it would, but at least you might want to display a message indicating such.
It aggregates market indicators that have been known to generate contrarian buy/sell signals. CNN Fear & Greed, Bank of America SSI, AAII Investor Sentiment Survey among others.
A few technical details:
* vibe coded ~70% of it — the parts that not were either UI polish that was faster to do directly or points of the data scraping pipeline where Claude got stuck
* architecture: vanilla JS frontend + Python/Flask backend + background jobs that run an AI data extraction pipeline (Perplexity Sonar / Exa for search + GPT-5)
* runs on render.com as web service + two cron jobs that run the data update process and send daily email notifications
At some point I did try Codex for a few PRs that were never merged and I re-did instead with Claude.