Hacker News Reader

Comment by tekacs | Hacker News Reader

I code with AI all day, every day. But I do think that it's worth pointing to this issue (from March).

The author has said that they've redone it since, but the "from-scratch hand-built" framing specifically – for me – somewhat grates given the original heavy lifting from an existing AGPL codebase.

https://github.com/cesanta/elk/issues/75

I want to acknowledge that the original authors don't seem to have minded too much – per that thread – after older versions were dropped.

For context, the current code doesn't look like it is the same shape, the same structure, etc., etc. – it _has_ been rewritten since (the 'since Feb' rewrite mentioned adjacent is related to this, AFAICT).

To the author: I absolutely love what you're doing overall. Keep going! Just be careful, folks.

View on HN

Ok, but this post: https://themackabu.dev/blog/js-in-one-month

isn't the post of someone who just implemented a js engine (it reads like someone who asked an LLM to write a blog post about the git log of a different LLM which was apparently lifting code from a different js engine...)

It's a bit hard to understand what's going on here, but definitely hard to trust the project.

Apparently prompting LLMs is now "survival mode".

this was flagging code from all the way back in dec of 2025, back when this project was just some idea and not what it is today.

around feb thats when basically deleted the existing codebase and designed a much more reliable system from the ground up.

I knew the existing code was basically pure slop, and it was not the biggest issue then, now nothing goes past me unreviewed and untested

> To the author: I absolutely love what you're doing overall. Keep going! Just be careful, folks.

Thank you!

> now nothing goes past me unreviewed and untested

So it's not hand-built?

OP, you will get a much better reception if you clarify that "hand built" just means you're not using existing libraries. It's a cool project, but it's disappointing to come in expecting to see a rare gem of a 100% human project in 2026 but it's really something else.

This 100%

Honest question - does using claude code and steering it to type the code I want counts as hand-built? Like doing actual PRs, nit picking, basically treating it as a jr/product/typist while I'm the one making ALL the decisions? Reading every line and approving manually if it's a code that I would write?

No it does not count as hand built. In the same way that a human pressing buttons, such that an automated production line makes a watch, and then at the end a human goes and looks at it and says "yup that's a watch" isn't a hand-made watch.

Nice opinion but it's clearly a spectrum and not binary. Is assembling machine-made components by hand to make a watch hand-made?

I don't call my Ikea furniture hand-built.

In the age of LLMs we can assume that any code was generated by an LLM unless noted otherwise, so I think it is fair to not point out the use of LLMs. However, words like hand built should be reserved for the instances where the code was actually written unassisted. Reading all the code means nothing, there is no such thing as code that was written as if you had written it, unless you have written it.

You're not making all the decisions in that case. You're making approvals and steering but the agent is making decisions and is framing what is even available to you to approve. When you build something by hand, you make decisions as you go in a way that the agent is now replacing.

Directly equivalent to having outsourced it to a shop of remote consultants where you remained as onshore lead. And in all those past cases we would have said "No."

What is?

I, at the very least, need a keyboard.

Does using non AI code autocompletions and code templates makes your code non hand written? Because I think it’s based on same idea - “prediction of next word”. And sometimes with just few buttons pressed you get surprisely good lines.

Unreviewed, untested, understood?

Reply on news.ycombinator.com