<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: scronkfinkle</title><link>https://news.ycombinator.com/user?id=scronkfinkle</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 18 Nov 2026 14:17:15 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=scronkfinkle" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[Show HN: A local-first AI agent for the desktop, no Electron, any LLM endpoint]]></title><description><![CDATA[
<p>Hey folks. Most all the apps I see for agent harnesses are typescript based, and involve either electron or some form of IPC to function as a desktop app.<p>In the age of LLM's, I thought it would be cool to build out a prototype of what a modern desktop application would like for this, without using any web code, since I can have agents pretty aggressively work on the cross-platform compatibility on my behalf. After about a week, I ended up with a harness that feels pretty good!<p>Feel free to try it from the releases or you can install directly via:<p>```   
cargo install byob   
````</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50507341">https://news.ycombinator.com/item?id=50507341</a></p>
<p>Points: 4</p>
<p># Comments: 0</p>
]]></description><pubDate>Sun, 15 Nov 2026 14:52:06 +0000</pubDate><link>https://github.com/Scronkfinkle/byob</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50507341</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50507341</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Show HN: Black Ops II statically recompiled to native ARM64 macOS"]]></title><description><![CDATA[
<p>> does not use any emulator whatsoever, be it Wine, Crossover or Rosetta<p>Not sure about the other two, but WINE Is Not an Emulator. It's in the name :)</p>
]]></description><pubDate>Wed, 11 Nov 2026 14:02:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=50453553</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50453553</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50453553</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Psalm 7 release: PHP security&static analysis software"]]></title><description><![CDATA[
<p>I dunno,unlike the crypto/NFT era, Rust rewrites have demonstrated measured performance and memory savings gains. Some grimace when they see "written in Rust" as a tagline but my immediate thought tends to land more on the "thank God this isn't an electron app" side of things</p>
]]></description><pubDate>Mon, 09 Nov 2026 12:51:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=50424899</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50424899</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50424899</guid></item><item><title><![CDATA[New comment by scronkfinkle in "The Grid, a factory-scale robot in Massachusetts that builds PCBs on demand"]]></title><description><![CDATA[
<p>Earlier this month I ordered some PCBs. Due to the tariffs OSH park was actually a viable competitor for small batch runs. I've also found them to be higher quality than PCB way</p>
]]></description><pubDate>Wed, 28 Oct 2026 17:09:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=50276923</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50276923</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50276923</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Git for the US Code"]]></title><description><![CDATA[
<p>Let's say you have some case law, and to know if it's relevant you need to pull the primary law/codes and regulations .You also need to know when the correct versions of everything are, so the case law needs to resolve to the correct version of the correct piece of legislation. "What is an atomic unit of legislation?" is also fuzzy as amendments will both move and edit named blocks of legislation at once. It's a big messy very hard to track pile of moving targets, and you need to account for the temporal aspect while linking documents not only to the changes but also to documents at certain revisions. Standard git and other version control has trouble expressing many of these concepts, and while it could be hacked in, I have found it more effective to create a purpose-built tool for the job.</p>
]]></description><pubDate>Wed, 28 Oct 2026 14:45:12 +0000</pubDate><link>https://news.ycombinator.com/item?id=50274537</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50274537</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50274537</guid></item><item><title><![CDATA[Show HN: Local, Tracked, Versioned, Legal Research Databases]]></title><description><![CDATA[
<p>This is a project I've been working on for about 8 months now. I worked in legal tech for 3 years and found that one of the biggest hurdles for AI agents and lawyers alike when doing legal research is tracking changes, and correctly cross referencing those changes between different forms of legal documents.<p>W2D builds a SQLite dataset from public sources, and offers agent skills for answering questions about that dataset.<p>The dataset holds multiple revisions of the U.S. Code at each release point that you add. It holds the public laws that amend the Code, and it links each amendment to the change that it made. It holds court opinions from CourtListener, by page and paragraph, and it links each U.S.C. citation to the provision that the court cited. It can also hold House roll calls.<p>This is similar in spirit to "git for Law" but includes a lot more utilities than diffs to make it an effective tool for research.</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=50274235">https://news.ycombinator.com/item?id=50274235</a></p>
<p>Points: 2</p>
<p># Comments: 0</p>
]]></description><pubDate>Wed, 28 Oct 2026 14:29:47 +0000</pubDate><link>https://www.wordstodata.com/</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50274235</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50274235</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Git for the US Code"]]></title><description><![CDATA[
<p>I've been working on this problem for several months now:  <a href="https://www.wordstodata.com/" rel="nofollow">https://www.wordstodata.com/</a><p>A surprising amount of tracking and diffing in legal texts is actually very different to git. It's a trap to assume software version control fits the needs for legal work.</p>
]]></description><pubDate>Wed, 28 Oct 2026 07:00:57 +0000</pubDate><link>https://news.ycombinator.com/item?id=50270056</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50270056</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50270056</guid></item><item><title><![CDATA[New comment by scronkfinkle in "NixOS is making me less afraid"]]></title><description><![CDATA[
<p>A lot of folks have been praising the harmony of Claude code + nixOS. For those of you doing this, I would recommend you to ask your agent to switch your config to nix flakes. It's a much better paradigm for working with the versioning than the default channels interface. Additionally, you should ask your agent to look into home manager, if it hasn't done so already.</p>
]]></description><pubDate>Tue, 20 Oct 2026 13:57:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=50166560</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50166560</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50166560</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Recreational Programming"]]></title><description><![CDATA[
<p>I feel you're missing the point. AI has no impact on recreational programming, it's about satisfying your personal curiousities and interests. I'm currently hand rolling a database project in rust because I find writing rust is fun. If you're pausing to ask yourself "what's the use if you can just use AI" then you may not have liked programming intrinsically to begin with, but instead the second order effects it gave you.</p>
]]></description><pubDate>Wed, 14 Oct 2026 02:05:04 +0000</pubDate><link>https://news.ycombinator.com/item?id=50081499</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=50081499</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=50081499</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Clef: Open-weight decision models, and new RL fine-tuning platform"]]></title><description><![CDATA[
<p>> no output price listed<p>It's weird to think of these kinds of models as having "output tokens". Cross-encoder approaches like Laya add a [MASK] marker per option, but nothing is generated the way an autoregressive transformer generates. It's one bidirectional pass over your input, then a small head scores each option, so you wouldn't really pay for output as much as only input</p>
]]></description><pubDate>Thu, 01 Oct 2026 20:30:03 +0000</pubDate><link>https://news.ycombinator.com/item?id=49926693</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49926693</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49926693</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Ollaya – Ollama for open-source, Jev-style decision models"]]></title><description><![CDATA[
<p>Yes. JEV generalizes better because they probably have an enormous corpus and trained on it for a long time. Laya's out of the box model is much weaker. However, in the age of LLM's it's incredibly easy and cheap to generate large datasets to fine tune laya for your task, and the training loop is pretty quick and cheap too.<p>It's so easy that I question why I would ever pay for JEV when eventually I'll have done enough random things that I will also have a large corpus and likely a general model as well.</p>
]]></description><pubDate>Fri, 25 Sep 2026 19:05:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=49848596</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49848596</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49848596</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Programming Tutorials Are Dead"]]></title><description><![CDATA[
<p>You're correct that writing code by hand has been heavily displaced by AI programming, but understanding how things work isn't really an "antiquated hobby" and this mentality causes a lot of people to incorrectly assume that latent knowledge of how to program is no longer useful.<p>It's still pretty undisputed that if you know how to write code, you can generate better AI code than someone who doesn't, and do it with less tokens. I think it's more like in university where in other engineering disciplines you learn the math and underlying concepts and are forced to do them by hand. This gives you an understanding of what you're doing, but in the real world you, ya know, just plug that stuff into matlab.</p>
]]></description><pubDate>Thu, 24 Sep 2026 20:09:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=49836168</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49836168</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49836168</guid></item><item><title><![CDATA[New comment by scronkfinkle in "PS5 Linux lead quits: "a bunch of noobs using LLMs" that "they don't understand""]]></title><description><![CDATA[
<p>> remember hacktoberfest 2020?<p>Oh man, if you read the threads about that it's such a time capsule of a different era:<p>e.g. from this thread: <a href="https://news.ycombinator.com/item?id=31628342">https://news.ycombinator.com/item?id=31628342</a><p>> I am honestly surprised how little SPAM there is on GitHub in general. Please don’t take that as a challenge!</p>
]]></description><pubDate>Wed, 16 Sep 2026 15:12:25 +0000</pubDate><link>https://news.ycombinator.com/item?id=49728294</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49728294</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49728294</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Measuring the sloppiness of code"]]></title><description><![CDATA[
<p>There is some sense of rose-tinted glasses of pre-LLM coding. A lot of human written code, particularly at the enterprise level, was of low quality well before AI automated it.</p>
]]></description><pubDate>Fri, 11 Sep 2026 14:21:13 +0000</pubDate><link>https://news.ycombinator.com/item?id=49658951</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49658951</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49658951</guid></item><item><title><![CDATA[New comment by scronkfinkle in "The Waymo effect: how AI is quietly making research less collaborative"]]></title><description><![CDATA[
<p>I wanted you to be wrong, and to be able to make this an example of us over-reacting to certain trigger words created by AI, but unfortunately I just scanned the first couple paragraphs with pangram and it reported 100% AI, so you're probably correct.<p>I get a funny feeling in my stomach over the idea that common and effective means of communication (i.e. it's not X it's Y) have become faux pas to use because of AI. I think it's something about these phrases being taken away from us more-so than the AI inventing them.</p>
]]></description><pubDate>Fri, 11 Sep 2026 11:50:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=49656852</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49656852</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49656852</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Claude is only available to people over 18 years"]]></title><description><![CDATA[
<p>That's a valid take. The issue I'm wrestling with is the inevitable attempts to point a finger at who is responsible when bad things happen. If you claim it is on the tech companies to know if a child is online, then they will take the safest path for them by forcing ID verification in independent adhoc manners. This is what we are seeing now, and to me this is the worst situation. If you shift the responsibility more to the parents (this child was using a device that didn't send the `PARENT_CONTROL` flag, therefore we assumed they were an adult) it's a completely different conversation. Furthermore, if something happens to a child AND it's obvious from traffic logs that the platform willingly knew a child was being talked to, then that is also a completely different situation than the first.<p>Leaving it all completely deregulated and/or letting platforms implement it themselves to varying levels of success and personal invasion feels like the worst option to me.</p>
]]></description><pubDate>Fri, 11 Sep 2026 11:45:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=49656794</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49656794</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49656794</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Claude is only available to people over 18 years"]]></title><description><![CDATA[
<p>I think you may be  misunderstanding a bit.  There's no forced verification. It'd be more of an RFC that gives parents the ability to communicate their underage child is using the device without revealing or verifying any further information. Or do you suspect that giving any ground will cause the ID verification and centralized behavior?</p>
]]></description><pubDate>Fri, 11 Sep 2026 11:28:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49656623</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49656623</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49656623</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Claude is only available to people over 18 years"]]></title><description><![CDATA[
<p>I once heard someone suggest that this should be on the OS level and I'm slowly coming around to the idea. There should be some kind of OS level flag that can easily broadcast to products that a child is using the device. No ID verification required, the parent is responsible for setting it, and all downstream applications are regulated to respect it, If applicable to the product. This removes the incredibly invasive ID verification, and places responsibility on the people responsible for the minor. Companies like meta then can be sued not for failing to detect children (this is, evidently, not working on any platform trying to enforce this currently), but instead can be judged in a black and white manner (i.e. is the child version of meta too predatory to children?)</p>
]]></description><pubDate>Fri, 11 Sep 2026 11:22:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=49656543</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49656543</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49656543</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra"]]></title><description><![CDATA[
<p>But I don't want to use your CLI. I already have my own harnesses and workflows. The friction is too high to "just try out" a new model like this. It would be preferable if I can evaluate it over, say, open router like all the other models and then decide from there if it's worth downloading a bespoke tool chain for only 1 lab's models</p>
]]></description><pubDate>Thu, 10 Sep 2026 16:44:54 +0000</pubDate><link>https://news.ycombinator.com/item?id=49646671</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49646671</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49646671</guid></item><item><title><![CDATA[New comment by scronkfinkle in "Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra"]]></title><description><![CDATA[
<p>Please correct me if I'm wrong, but this appears to require Devin to use? I'm disappointed to see I need to use a bespoke platform to interact with this agent, to the point that I probably won't be trying it.</p>
]]></description><pubDate>Thu, 10 Sep 2026 16:13:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=49646173</link><dc:creator>scronkfinkle</dc:creator><comments>https://news.ycombinator.com/item?id=49646173</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49646173</guid></item></channel></rss>