Show HN: Action Engine — An API/Agent Buildkit Putting Flexibility First https://ift.tt/R84lmQ3

Show HN: Action Engine — An API/Agent Buildkit Putting Flexibility First https://ift.tt/84pGkWA October 28, 2025 at 01:56AM

Show HN: JSON Query https://ift.tt/LVXKpnZ

Show HN: JSON Query I'm working on a tool that will probably involve querying JSON documents and I'm asking myself how to expose that functionality to my users. I like the power of `jq` and the fact that LLMs are proficient at it, but I find it right out impossible to come up with the right `jq` incantations myself. Has anyone here been in a similar situation? Which tool / language did you end up exposing to your users? https://ift.tt/ilJcHBP October 27, 2025 at 09:52PM

Show HN: Helium Browser for Android with extensions support, based on Vanadium https://ift.tt/stR3vCD

Show HN: Helium Browser for Android with extensions support, based on Vanadium Been working on an experimental Chromium-based browser that brings 2 major features to your phone/tablet: 1. desktop-style extensions: natively install any extensions (like uBO) from the chrome web store, just toggle "desktop site" in the menu first. 2. privacy/security hardening: applies the full patch sets from Vanadium (with Helium's currently wip). Means you get both browsers' excellent privacy features, like Vanadium's webrtc IP policy option that protects your real IP by default, and security improvements such as JIT being disabled by default, all while being a reasonably efficient FOSS app that can be installed on any (modern) android. It's still in beta, and as I note in the README, it's not a replacement for the full OS-level security model you'd get from running the GrapheneOS Vanadium combo. However, goal was to combine privacy of Vanadium with the power of desktop extensions and Helium features, and make it accessible to a wider audience. (Passkeys from Bitwarden Mobile should also work straight away once merged in the list of FIDO2 privileged browsers) Build scripts are in the repo if you want to compile it yourself. You can find pre-built releases there too. Would love any feedback/support! https://ift.tt/Yw7I5DF October 27, 2025 at 04:11AM

Show HN: The Legal Embedding Benchmark (MLEB) https://ift.tt/DJf5cBK

Show HN: The Legal Embedding Benchmark (MLEB) Hey HN, I'm excited to share the Massive Legal Embedding Benchmark (MLEB) — the first comprehensive benchmark for legal embedding models. Unlike previous legal retrieval datasets, MLEB was created by someone with actual domain expertise (I have a law degree and previously led the AI team at the Attorney-General's Department of Australia). I came up with MLEB while trying to train my own state-of-the-art legal embedding model. I found that there were no good benchmarks for legal information retrieval to evaluate my model on. That led me down a months-long process working alongside my brother to identify or, in many cases, build our own high-quality legal evaluation sets. The final product was 10 datasets spanning multiple jurisdictions (the US, UK, Australia, Singapore, and Ireland), document types (cases, laws, regulations, contracts, and textbooks), and problem types (retrieval, zero-shot classification, and QA), all of which have been vetted for quality, diversity, and utility. For a model to do well at MLEB, it needs to have both extensive legal domain knowledge and strong legal reasoning skills. That is deliberate — given just how important high-quality embeddings are to legal RAG (particularly for reducing hallucinations), we wanted our benchmark to correlate as strongly as possible with real-world usefulness. The dataset we are most proud of is called Australian Tax Guidance Retrieval. It pairs real-life tax questions posed by Australian taxpayers with relevant Australian Government guidance and policy documents. We constructed the dataset by sourcing questions from the Australian Taxation Office's community forum, where Australian taxpayers ask accountants and ATO officials their tax questions. We found that, in most cases, such questions can be answered by reference to government web pages that, for whatever reason, users were unable to find themselves. Accordingly, we manually went through a stratified sample of 112 challenging forum questions and extracted relevant portions of government guidance materials linked to by tax experts that we verified to be correct. What makes the dataset so valuable is that, unlike the vast majority of legal information retrieval evaluation sets currently available, it consists of genuinely challenging real-world user-created questions, rather than artificially constructed queries that, at times, diverge considerably from the types of tasks embedding models are actually used for. Australian Tax Guidance Retrieval is just one of several other evaluation sets that we painstakingly constructed ourselves simply because there weren't any other options. We've contributed everything, including the code used to evaluate models on MLEB, back to the open-source community. Our hope is that MLEB and the datasets within it will hold value long into the future so that others training legal information retrieval models won't have to detour into building their own "MTEB for law". If you'd like to head straight to the leaderboard instead of reading our full announcement, you can find it here: https://ift.tt/9SgvqwO If you're interested in playing around with our model, which happens to be ranked first on MLEB as of 16 October 2025 at least, check out our docs: https://ift.tt/T9fryLB https://ift.tt/ukoseWP October 27, 2025 at 03:46AM

Show HN: MyraOS – My 32-bit operating system in C and ASM (Hack Club project) https://ift.tt/Yc0p8bK

Show HN: MyraOS – My 32-bit operating system in C and ASM (Hack Club project) Hi HN, I’m Dvir, a young developer. Last year, I got rejected after a job interview because I lacked some CPU knowledge. After that, I decided to deepen my understanding in the low level world and learn how things work under the hood. I decided to try and create an OS in C and ASM as a way to broaden my knowledge in this area. This took me on the most interesting ride, where I’ve learned about OS theory and low level programming on a whole new level. I’ve spent hours upon hours, blood and tears, reading different OS theory blogs, learning low level concepts, debugging, testing and working on this project. I started by reading University books and online blogs, while also watching videos. Some sources that helped me out were OSDev Wiki ( https://ift.tt/BdNKV8W ), OSTEP ( https://pages.cs.wisc.edu/~remzi/OSTEP ), open-source repositories like MellOS and LemonOS (more advanced), DoomGeneric, and some friends that have built an OS before. This part was the longest, but also the easiest. I felt like I understood the theory, but still could not connect it into actual code. Sitting down and starting to code was difficult, but I knew that was the next step I needed to take! I began by working on the bootloader, which is optional since you can use a pre-made one (I switched to GRUB later), but implementing it was mainly for learning purposes and to warm up on ASM. These were my steps after that: 1) I started implementing the VGA driver, which gave me the ability to display text. 2) Interrupts - IDT, ISR, IRQ, which signal to the CPU that a certain event occurred and needs handling (such as faults, hardware connected device actions, etc). 3) Keyboard driver, which enables me to display the same text I type on my keyboard. 4) PMM (Physical memory management) 5) Paging and virtual memory management 6) RTC driver - clock addition (which was, in my opinion, optional) 7) PIT driver - Ticks every certain amount of time, and also 8) FS (File System) and physical HDD drivers - for the HDD I chose PATA (HDD communication protocol) for simplicity (SATA is a newer but harder option as well). For the FS I chose EXT2 (The Second Extended FileSystem), which is a foundational linux FS structure introduced in 1993. This FS structure is not the simplest, but is very popular in hobby-OS, it is very supported, easy to set up and upgrade to newer EXT versions, it has a lot of materials online, compared to other options. This was probably the longest and largest feature I had worked on. 9) Syscall support. 10) Libc implementation. 11) Processing and scheduling for multiprocessing. 12) Here I also made a shell to test it all. At this point, I had a working shell, but later decided to go further and add a GUI! I was working on the FS (stage 8), when I heard about Hack Club’s Summer of Making (SoM). This was my first time practicing in HackClub, and I want to express my gratitude and share my enjoyment of participating in it. At first I just wanted to declare the OS as finished after completing the FS, and a bit of other drivers, but because of SoM my perspective was changed completely. Because of the competition, I started to think that I needed to ship a complete OS, with processing, GUI and the bare minimum ability to run Doom. I wanted to show the community in SoM how everything works. Then I worked on it for another 2 months, after finishing the shell, just because of SoM!, totalling my project to almost 7 months of work. At this time I added full GUI support, with dirty rectangles and double buffering, I made a GUI mouse driver, and even made a full Doom port! things I would've never even thought about without participating in SoM. This is my SoM project: https://ift.tt/0rMqIN5 . Every project has challenges, especially in such a low level project. I had to do a lot of debugging while working on this, and it is no easy task. I highly recommend using GDB which helped me debug so many of my problems, especially memory ones. The first major challenge I encountered was during the coding of processes - I realized that a lot of my paging code was completely wrong, poorly tested, and had to be reworked. During this time I was already in the competition and it was difficult keeping up with devlogs and new features while fixing old problems in a code I wrote a few months ago. Some more major problems occurred when trying to run Doom, and unlike the last problem, this was a disaster. I had random PFs and memory problems, one run could work while the next one wouldn’t, and the worst part is that it was only on the Doom, and not on processes I created myself. These issues took a lot of time to figure out. I began to question the Doom code itself, and even thought about giving up on the whole project. After a lot of time spent debugging, I fixed the issues. It was a combination of scheduling issues, Libc issues and the Qemu not having enough (wrongfully assuming 128MB for the whole OS was enough). Finally, I worked throughout all the difficulties, and shipped the project! In the end, the experience working on this project was amazing. I learned a lot, grew and improved as a developer, and I thank SoM for helping to increase my motivation and make the project memorable and unique like I never imagined it would be. The repo is at https://ift.tt/dNvIc24 . I’d love to discuss any aspect of this with you all in the comments! https://ift.tt/dNvIc24 October 27, 2025 at 02:13AM

Show HN: I Built DevTools for Blazor (Like React DevTools but for .NET) https://ift.tt/gLuqbHm

Show HN: I Built DevTools for Blazor (Like React DevTools but for .NET) Hi HN! I've been working on developer tools for Blazor that let you inspect Razor components in the browser, similar to React DevTools or Vue DevTools. The problem: Blazor is Microsoft's frontend framework that lets you write web UIs in C#. It's growing fast but lacks the debugging tools other frameworks have. When your component tree gets complex, you're stuck with Console.WriteLine debugging. What I built: A browser extension + NuGet package that: Shows the Razor component tree in your browser Maps DOM elements back to their source components Highlights components on hover Works with both Blazor Server and WASM How it works: The NuGet package creates shadow copies of your .razor files and injects invisible markers during compilation. These markers survive the Razor→HTML pipeline. The browser extension reads these markers to reconstruct the component tree. Current status: Beta - it works but has rough edges. Found some bugs when testing on larger production apps that I'm working through. All documented on GitHub. Technical challenges solved: Getting markers through the Razor compiler without breaking anything Working around CSS isolation that strips unknown attributes Making it work with both hosting models It's completely open source: https://ift.tt/OnR0TAa Demo site where you can try it: https://ift.tt/QbMJkAB Would love feedback, especially from anyone building production Blazor apps. What debugging pain points do you have that developer tools could solve? https://ift.tt/GtjA7Cp October 26, 2025 at 10:04PM

Show HN: I created a small 2D game about an ant https://ift.tt/mG1xUjR

Show HN: I created a small 2D game about an ant Hello HN! I created a short game in just a few days, just for fun, where you play as an ant and feed it apples. This game also features random landscape generation, where clouds and trees are randomly distributed across all coordinates (only trees do not grow in the y direction). This is what took me the longest time :) I would appreciate your feedback ^ ^ https://ift.tt/87xT0zg October 26, 2025 at 12:50AM

Show HN: Random Makers – Show HN and Product Hunt, but Faster and Not Corporate https://ift.tt/GxwE571

Show HN: Random Makers – Show HN and Product Hunt, but Faster and Not Corporate https://ift.tt/x83q4rK October 25, 2025 at 11:32PM

Show HN: Wsgrok – one of many ngrok alternatives https://ift.tt/HpLmoeC

Show HN: Wsgrok – one of many ngrok alternatives I built it for myself because ngrok didn't let me add one more domain unless I paid $12 more. I probably should've looked for alternatives before building my own, but my grudge got in the way . Once I started, I wanted to make it better than the other options. Silly, I know. No one probably cares. I plan to open source it sometime next year because I’ve got other projects to finish first. It's free until I deplete my cloud credits, then it will switch to a tier-based model with a free option. https://wsgrok.com October 25, 2025 at 05:16AM

Show HN: Pensive – A bookmarking tool with full-text search and LLMs https://ift.tt/UpsYaMI

Show HN: Pensive – A bookmarking tool with full-text search and LLMs After Pocket shut down, I started working on a new bookmarking solution. My main goal was to make bookmarks searchable, something that Pocket was not good at. So I built Pensive, a bookmarking solution that saves the full page content and makes it searchable with full-text search. You can add pages using a browser extension or a Telegram bot (for saving on mobile). It is written in Go with PostgreSQL, all Dockerized on a $5 Hetzner server, and the front end is built with HTMX. I have also added embeddings (using Gemini Flash Lite) to let LLMs interact with your bookmarks contextually. It is stable enough that I now use it daily. I’m considering publishing it as open source, but first I want to have a proper version ready. Feel free to email me if you’re interested in helping out or if you have prior experience with open source. Feel free to try it at https://getpensive.com https://getpensive.com/ October 25, 2025 at 03:28AM

Show HN: The Σ-Manifold Manifesto https://ift.tt/D42x5sz

Show HN: The Σ-Manifold Manifesto This project explores the connection between *the linear structure of text* and its *emotional-aesthetic impact*. We identify *five fundamental relations* between consecutive sentences — labeled *A–E*. Each represents a shift of *subject-object*, i.e., a transformation of perspective and agency. When texts grow longer, these relations form *sequences* — and from the infinite combinatorial space, *eight stable patterns (Σ₁–Σ₈)* emerge empirically. Each pattern correlates with a distinct *semantic and emotional field* — cathartic, heroic, meditative, humorous, and so on. This allows us to instruct an LLM not through semantic prompts (“write a story about…”), but through *structural commands* — e.g., generate a narrative following sequence Σ₅ (Tragic Counterpoint). You can experiment with these archetypes directly here: [Narrative Generator]( https://ift.tt/SmRZya7... ) or [via python]( https://ift.tt/ULfv684... ) Interestingly, there appears to be a parallel between these textual progressions and *musical harmony*. For example, if A–E are mapped to harmonic functions (I, IV, V, vi, ii), the narrative sequences behave like emotional “chord progressions” — where meaning flows, modulates, and resolves. Coherence in the generated text arises not only from syntax, but from the *associative field* that the LLM constructs around these shifting relations. When asked to “switch subjects,” it spontaneously moves from Poet to Writer , preserving aesthetic continuity rather than randomness. It might even hint at how *children acquire language*: by first sensing the melody of structural transitions, before mapping them to concepts and emotions. Such a method could eventually apply to *training neural systems*, where meaning is learned as flow — not as fixed representation. [Full text]( https://ift.tt/ockTvpx... ) October 25, 2025 at 01:34AM

Show HN: Check what is hogging your disk zpace https://ift.tt/BzojHZX

Show HN: Check what is hogging your disk zpace Hello world! I would like to share that I have created a simple open-source Python CLI app to check what's hogging all the disk space! You can install it with pip. It's like space but Zpace. pip install zpace and find the big files consuming your disk space. Be it apps, virtual environments, machine learning models etc. Just run zpace Once you find one, that you can say "bye" to, just run rm -rf /i_dont_need/this/file/right_now to get rid of it. It was born out of frustration while lack of disk space prevented me to use my laptop properly. It has been very useful to me so far. I hope that it can be useful to you as well. Feel free to check it out. Currently tested on MacOS only https://ift.tt/CkN7Vwt October 25, 2025 at 01:12AM

Show HN: Git for LLMs – a context management interface https://ift.tt/VSmGoYW

Show HN: Git for LLMs – a context management interface Hi HN, we’re Jamie and Matti, co-founders of Twigg. During our master’s we continually found the same pain points cropping up when using LLMs. The linear nature of typical LLMs interfaces - like ChatGPT and Claude - made it really easy to get lost without any easy way to visualise or navigate your project. Worst of all, none of them are well suited for long term projects. We found ourselves spending days using the same chat, only for it to eventually break. Transferring context from one chat to another is also cumbersome. We decided to build something more intuitive to the ways humans think. We started with two simple ideas. Enabling chat branching for exploring tangents, and an interactive tree diagram to allow for easy visualisation and navigation of your project. Twigg has developed into an interface for context management - like “Git for LLMs”. We believe the input to a model - or the context - is fundamental to its performance. To extract the maximum potential of an LLM, we believe the users need complete control over exactly what context is provided to the model, which you can do using simple features like cut, copy and delete to manipulate your tree. Through Twigg, you can access a variety of LLMs from all the major providers, like ChatGPT, Gemini, Claude, and Grok. Aside from a standard tiered subscription model (free, plus, pro), we also offer a Bring Your Own Key (BYOK) service, where you can plug and play with your own API keys. Our target audience are technical users who use LLMs for large projects on a regular basis. If this sounds like you, please try out Twigg, you can sign up for free at https://twigg.ai/ . We would love to get your feedback! https://twigg.ai October 23, 2025 at 08:42PM

Show HN: Tommy – Turn ESP32 devices into through-wall motion sensors https://ift.tt/LKxYT5N

Show HN: Tommy – Turn ESP32 devices into through-wall motion sensors Hi HN! I would like to present my project called TOMMY, which turns ESP32 devices into motion sensors that work through walls and obstacles using Wi-Fi sensing. TOMMY started as a project for my own use. I was frustrated with motion sensors that didn't detect stationary presence and left dead zones everywhere. Presence sensors existed but were expensive and needed one per room. I explored echo localization first, but microphones listening 24/7 felt too creepy. Then I discovered Wi-Fi sensing - a huge research topic but nothing production-ready yet. It ticked all the boxes: could theoretically detect stationary presence through breathing/micromovements and worked through walls and furniture so devices could be hidden away. Two years and dozens of research papers later, TOMMY has evolved into software I'm honestly quite proud of. Although it doesn't have stationary presence detection yet (coming Q1 2026) it detects motion really well. It works as a Home Assistant Add-on or Docker container, supports a range of ESP32 devices, and can be flashed through the built-in tool or used alongside existing ESPHome setups. I released the first version a couple of months ago on Home Assistant's subreddit and got a lot of interest and positive feedback. More than 200 people joined the Discord community and almost 2,000 downloaded it. Right now TOMMY is in beta, which is completely free for everyone to use. I'm also offering free lifetime licenses to every beta user who joins the Discord channel. You can read more about the project on https://ift.tt/kuWj0E2 . Please join the Discord channel if you are interested in the project. A note on open source: There's been a lot of interest in having TOMMY as an open source project, which I fully understand. I'm reluctant to open source before reaching sustainability, as I'd love to work on this full time. However, privacy is verifiable - it's 100% local with no data collection (easily confirmed via packet sniffing or network isolation). Happy to help anyone verify this. https://ift.tt/kuWj0E2 October 23, 2025 at 10:34PM

Show HN: Middlerok – reduces front end-back end integration from weeks to hours https://ift.tt/oA7lGJj

Show HN: Middlerok – reduces front end-back end integration from weeks to hours Generate production-ready OpenAPI specs, frontend & backend code and documentation with AI https://ift.tt/SWoHZNG October 22, 2025 at 11:05PM

Show HN: Incremental JSON parser for streaming LLM tool calls in Ruby https://ift.tt/GUnFxMq

Show HN: Incremental JSON parser for streaming LLM tool calls in Ruby Built this for streaming AI tool calls. LLMs stream function arguments as JSON character-by-character. Most parsers reparse from scratch each time - O(n²) behavior that causes UI lag. This maintains parsing state, processing only new characters. True O(n) performance that stays imperceptible throughout the entire response. Ruby gem, MIT licensed. Would love feedback. https://ift.tt/HntQT5R October 23, 2025 at 01:02AM

Show HN: I use ChatGPT these days to develop new features quickly https://ift.tt/KDhS0vo

Show HN: I use ChatGPT these days to develop new features quickly https://ift.tt/4CyfwzE October 22, 2025 at 02:28AM

Show HN: MTOR – A free, local-first PWA to automate workout progression https://ift.tt/ElvjZ6f

Show HN: MTOR – A free, local-first PWA to automate workout progression Hi HN, My motivation for this came from frustration with existing workout trackers. Most felt clunky, hid core features like performance graphs behind a paywall, or forced a native app download. A few people close to me who take their training seriously shared the same sentiment, so I decided to build my own. I'm working on mTOR, a free, science-based workout tracker I built to automate progressive overload. It's a local-first PWA that works completely offline, syncs encrypted between your devices using passwordless passkeys, and allows for plan sharing via a simple link. The core idea is to make progression easier to track and follow. After a workout, it analyzes your performance (weight, reps, and RIR), highlights new personal records (PRs), and generates specific targets for your next session. It also reviews your entire program to provide scientific analysis on weekly volume, frequency, and recovery for each muscle group. This gets displayed visually on an anatomy model to help you learn which muscles are involved, and you can track your gains over time with historical performance charts for each exercise. During a workout, you get a total session timer, an automatic rest timer, and can see your performance from the last session for a clear target to beat. It automatically advances to the next incomplete exercise, and when you need to swap an exercise, it provides context-aware alternatives targeting the same muscles. It's also deeply customizable: * The UI has a dark theme, supports multiple languages (English, Spanish, German), lets you adjust the UI scale, and toggle the visibility of detailed muscle names, exercise types, historical performance badges, and a full history card. * You can set global defaults for weight units (kg/lbs), rest times, and plan targets, or enable/disable metrics like Reps in Reserve (RIR) and estimated 1-Rep Max. The exercise library can be filtered by your available equipment, you can create your own custom exercises with global notes, and there's a built-in weight plate calculator. * The progression system lets you define default rep ranges and RIR targets, or create specific overrides for different lifts (e.g., a 3-5 rep range for strength, 10-15 for accessories). * Editing is flexible: you can drag-and-drop to reorder days, exercises, and sets, duplicate workout days, track unilateral exercises (left/right side), and enter data with a quick wheel picker. I'll be here all day to answer questions. I'm also thinking about making the project open-source down the line and would be curious to hear any thoughts on that. Thanks for checking it out! https://mtor.club/ October 22, 2025 at 12:04AM

Show HN: bbcli – A TUI and CLI to browse BBC News like a hacker https://ift.tt/Yk7H5Rp

Show HN: bbcli – A TUI and CLI to browse BBC News like a hacker hey hn! I (re)built this TUI tool for browsing BBC News in the terminal, it uses an RSS feed for getting headlines and previews and you can read articles too. Try it out and let me know what you think! :) https://ift.tt/ikZK4Yl October 19, 2025 at 04:28PM

Show HN: Online Sourcerer – The best answer to 'source?' https://ift.tt/LtArB0x

Show HN: Online Sourcerer – The best answer to 'source?' Hello, I made this site to combat misinformation on the internet by allowing users to prove that their claim is valid by linking multiple sources and combining them in a single link It's very early stage so I would love feedback on: - What types of claims would be most useful to you? - How can I make verification/sourcing more robust? - Any features that would make this actually useful vs just interesting? Thanks in advance, feel free to roast :) https://ift.tt/IHXR6q2 October 21, 2025 at 03:46AM