This is basically my version of "what all could you throw into Postgres?"

My problem with vibe coding/LLM assisted engineering is that it's hard to get the basic stuff that is independent of the application itself, correct, so I just use this and make sure everything I build has some consistency.

You can pop this in and use it as the base for your app and add login, permissions, etc. quite cleanly.

Comments URL: https://news.ycombinator.com/item?id=48887121

Points: 4

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=48004924

Points: 4

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=47999846

Points: 2

# Comments: 1

I became a father recently (:D) and it's been an emotional rollercoaster for me. I had been frantically Googling my "symptoms" and asking around what's wrong with me, because it seems I've been quite sensitive since the birth of my baby.

One way to explain this is the Gordon Ramsay meme (https://imgflip.com/memetemplate/211147137/Oh-dear-dear-gorg..., LHS = my reaction to my baby, RHS = my reaction to other kids before my baby was born).

I think the article is spot on — the more time you spend with your baby and care for them, the more oxytocin you get and the more your testosterone drops (I cried when my baby first spoke — cooed, really — to me, for example, and that's just one instance).

Edit: I want to take this opportunity to say — fuck companies that don't give paternity leave. This is fucking hard to do alone, so be nice to your employees and offer paternity benefits. I'm in India, where paternity leave isn't required, so I was told to fuck off when I asked for time off.

"0.05% of domains" is a vanity metric -- what matters is how many requests were mis-served cross-user. "Cache-Control was respected where provided" is technically true but misleading when most apps don't set it because CDN was off. The status page is more honest here too: they confirmed content without cache-control was cached.

They call it a "trust boundary violation" in the last line but the rest of the post reads like a press release. No accounting of what data was actually exposed.

[1] https://status.railway.com/incident/X0Q39H56

I've been using Claude Code to spin up apps quickly, and I kept needing the same infrastructure every time - user auth, permissions, usage tracking, job queues. So I pulled it all into one SQL package that lives in Postgres. Now when I start a new app I just tell Claude to use Postkit and all that stuff is already there, no external services to set up. I can focus on the actual product and iterate fast.

It was also a good excuse to actually use stuff I'd studied for system design interviews - Zanzibar-style ReBAC for permissions, a double-entry ledger for usage metering, transactional job queues with SKIP LOCKED. ~15k lines of SQL across five modules, with a Python SDK. The SQL works from any language though.

https://github.com/varunchopra/postkit

Comments URL: https://news.ycombinator.com/item?id=46724580

Points: 3

# Comments: 0

I'm a bit of a skeptic. Claude Code is good, but I've had varied results during my usage. Even just 5 minutes ago, I asked CC to view the most recent commit diff using git show. Even when I provided the command, it was doing dumb shit like git show --stat and then running wc for some reason...

I've been working on something called postkit[1], which has required me to build incrementally on a codebase that started from nothing and has now grown quite a lot. As it's grown, Claude Code's performance has definitely dipped.

[1] https://github.com/varunchopra/postkit

Comments URL: https://news.ycombinator.com/item?id=45693055

Points: 3

# Comments: 0

CEOs, AI "thought leaders," and VCs are advertising LLMs as magic, and tools like v0 and Lovable as the next big thing. Every response from leaders is some variation of https://www.youtube.com/watch?v=w61d-NBqafM

On the ground, we know that creating CLAUDE.md or cursorrules basically does nothing. It’s up to the LLM to follow instructions, and it does so based on RNG as far as I can tell. I have very simple, basic rules set up that are never followed. This leads me to believe everyone posting on that thread on Cursor is an amateur.

Beyond this, if you’re working on novel code, LLMs are absolutely horrible at doing anything. A lot of assumptions are made, non-existent libraries are used, and agents are just great at using tokens to generate no tangible result whatsoever.

I’m at a stage where I use LLMs the same way I would use speech-to-text (code) - telling the LLM exactly what I want, what files it should consider, and it adds _some_ value by thinking of edge cases I might’ve missed, best practices I’m unaware of, and writing better grammar than I do.

Edit:

[1] To add to this, any time you use search or Perplexity or what have you, the results come from all this marketing garbage being pumped into the internet by marketing teams.

So glad to come across it again!

Somehow it's gotten worse since then, and I'm back to using Claude for serious work.

Gemini is like that guy who keeps talking but has no idea what he's actually talking about.

I still use Gemini for brainstorming, though I take its suggestions with several grains of salt. It's also useful for generating prompts that I can then refine and use with Claude.

Example from the comments...

> Great question. I haven’t found a perfect tool. A combination of Figma and Storybook and Zeroheight do a good job pulling it all together. But the real magic is the designer-developer relationship that stay in lock step. Those relationships are what it takes to be in tight sync.

_But the real magic is the designer-developer relationship that stay in lock step. Those relationships are what it takes to be in tight sync._

Who writes like this?

Comments URL: https://news.ycombinator.com/item?id=43137621

Points: 1

# Comments: 0

That's really neat! Thank you for answering my questions and all the best with your launch!