steveharrison

Love the idea, but would help trustworthiness if the design looked a little less vibe-coded.

Particularly excited to see reverse image search get added to real estate websites. I think ability to filter listings by aesthetic will save a lot of time.

Cool use-case!

I hadn't thought about this angle as to why WebKit hasn't implemented this, but yeah 100%.

Nice one!

I'm really excited about window.showDirectoryPicker and the local-first web apps it will enable. There's lots of talk about local-first sync engines, but the best sync engine is one you don't even manage, like the user's file system / cloud storage service!

This was on my work computer.

Haha love it. Would love to see the IT guy's face handing that back in.

Interesting to see whether their data provides a moat or not. Would be great to remove flakey FE selectors but want to see it in action first…

I wrote an article about improving my productivity as a programmer. Keen to hear your thoughts and if there's anything I've missed!

I like those video effects on Apple landing pages where they map scroll position to the position in a video, so asked Claude Code to generate a working example using the Apple video: https://www.apple.com/105/media/us/macbook-neo/2026/eee281c9...

Yes, I’m planning for the app itself to be subscription and/or one-time payment, but of course the database is all yours if you stop paying.

I haven’t used double-entry because I wanted to focus on spending analysis rather than account balances, and often bank statements are partial data. Would this be a deal-breaker for you?

https://steveharrison.dev - I post about web dev stuff usually! Tech stack: Astro hosted on GitHub Pages

I agree with the other things though like having multiple inputs for date ranges / trying to use native elements with just some custom styling.

I think this article misses the fact that the native date and time pickers look ugly in Chrome, and that a lot of websites are ultimately an extension of a brand, not just a tool where function > form. The Airbnb date range picker looking on-brand makes the experience seem a lot more slick. There are more things to optimise for than just accessibility.

This is really cool! I’ve always wanted to do something similar, but didn’t want to invest the time, so thanks for seeing this through!

Some nice thoughts on designing a beautiful product!

I'm working on yet another spending analysis app. How will this one be different?

- Stores transactions locally in a SQLite database, so in theory you could build your own Front End if mine didn't do everything you wanted. Think someone who wants to do a fancy "year end summary" visualisation. And if you want to do a complex filter, can just write an SQL query to the database itself.

- Has a great UI.

- Allows you to write your own import connector and mix 'n' mash between locally-saved CSV/OFX files and APIs, rather than going down the standard route of only supporting Plaid. Current solutions leave users high and dry if the Plaid connector doesn't work well / they want to import old data (Plaid only gives the last 2 years I think).

I'm building an app to do this, will let you know when it's in beta. :) Meanwhile, some of the similar apps that store data locally I've come across are: - https://copilot.money/ (nice UI; partly local—think it might use CloudKit for syncing) - https://ufincs.com/ - https://actualbudget.org/

I feel like http://monarchmoney.com is still the best UI I've come across, but I can't believe how common it is for all these providers to store your transactions in their cloud... seems like a fundamental security flaw to me, no matter how much they claim to lock down access to customer support agents/etc.