Show HN: KillSwitch – a programming language designed to be difficult for LLMs https://ift.tt/sTpo785

Show HN: KillSwitch – a programming language designed to be difficult for LLMs https://ift.tt/c2kYswM September 20, 2026 at 01:10AM

Show HN: CUA-S1 – A System One Model for Computer Use https://ift.tt/bH2Ygwk

Show HN: CUA-S1 – A System One Model for Computer Use Hello HN! We're Dillon and Francesco from Cua. We were wondering how many computer use tasks actually need a full general purpose LLM (e.g. gpt-6-astra, claude-opus-5 etc.) to think through all their decisions and steps. Some tasks require thinking about a plan, exploring different paths, recovering from failure. Other tasks are a question of making local decisions, like this value should go in this box, or should I check this box, or this element should be ignored. We wondered how far we could go with a small model trained to only make these kinds of decisions. Our inspiration was Typesafe's Jev and its System One Model framing. This is a nod to the dichotomy between thinking quickly, automatically, and intuitively (system 1) vs. thinking slowly, analytically (system 2), as described by Daniel Kahneman. The interesting question for us was: what happens if you give a model an interface of current context, and a set of possible choices, and you ask it to return a probability for each choice? This kind of model does not generate output token by token like most LLMs do, but rather scores the options you give it, which you can check, trust, and use to drive your app's behavior. CUA-S1 is our answer for narrow, specialized decision models for computer use. Our first release is CUA-S1-FORMS. We built this from ideas and code in jevlike, and then trained a second model just to handle form interactions. It has 706k parameters, and the original checkpoint is 2.8 MB. The first training iteration took less than 30 minutes on synthetic data. Given a set of structured elements and values extracted from a document, it predicts whether to use the given value, CHECK, CLICK, or SKIP for each element. It does not predict new values for text fields, and does not consider screenshots. Element decisions are scored together, and your code can order the actions, and Cua Driver will execute them one at a time. A first evaluation of this specialist vs. hosted Jev on our form task: - For the whole decision set: 99.7% correct vs 83.6%. - For the subset of steps that require an action: 100% correct vs 96%. - For the subset of steps that are just leaving already-filled fields alone: 100% correct vs 74%. The specialist was trained specifically for this task and convention (just press skip for already filled boxes), while hosted Jev has not been fine-tuned for it, so this is an experiment in scoped specialization. We measured 7-9 ms to score a form locally vs. 260-280 ms per call to hosted Jev including network latency, though those samples measure different things and are not end-to-end form completion times. Our interest here is in the space between a brittle script and a general agent loop. The content and layout of form fields vary enough that scripts get unwieldy, but the set of available decisions can remain narrow and well scoped. We want to explore the possibility of a general agent encountering something novel, and passing well understood decisions over to specialists like this. That is a direction we are looking into. The current release is for forms only. We're open sourced the synthetic data generation, training, evaluation, and Driver integration under libs/cua-s1 with an MIT license. Comments welcome! Especially if you are building computer-use agents and have run into a recurring decision that is too variable to script but is too narrow to call another LLM for. https://ift.tt/TiZVNdt September 19, 2026 at 09:22PM

Show HN: A model index from the AI Gateways https://ift.tt/xO7B0VR

Show HN: A model index from the AI Gateways https://ift.tt/TZzvA97 September 19, 2026 at 12:26AM

Show HN: Koi Editor Alpha https://ift.tt/pAZmsGV

Show HN: Koi Editor Alpha Hi, Koi Editor Alpha is now available to download for macOS 13.2+ (Apple Silicon). You can download the latest build here: https://ift.tt/r4jNRHw If you try it, feedback and bug reports are very welcome. You can also join the Koi Discord here: https://ift.tt/wXkQstP Thanks! -- Michael https://koieditor.com/ September 18, 2026 at 09:51PM

Show HN: Composing domain-specific harness on Python in 10 mins https://ift.tt/pFqAEY3

Show HN: Composing domain-specific harness on Python in 10 mins https://ift.tt/uOmbGie September 18, 2026 at 02:16AM

Show HN: The Endless Museum, Wikipedia as a walkable museum https://ift.tt/yInMvNR

Show HN: The Endless Museum, Wikipedia as a walkable museum https://ift.tt/sGofwkz September 18, 2026 at 12:28AM

Show HN: Ax-Check.com – Can Agents Use Your Product? https://ift.tt/jodp6bs

Show HN: Ax-Check.com – Can Agents Use Your Product? This is a free site that checks whether agents can smoothly navigate a site and try out the product. Agents can use this too, just point them to ax-check.com https://ift.tt/nE80hLJ September 17, 2026 at 11:38PM