🛠️ Hacker News Tools

[CLI Tool] Open Code Review – An AI-powered code review CLI tool

Man-Computer Symbiosis J. C. R. Licklider (1960)

Introducing Boron Buckyballs: Theory that B80 cages can’t be made is disproved

[Other] Show HN: Formally verified polygon intersection – Opus 4.8 oneshots, prev failed To my knowledge, this is the first formally verified implementation of an intersection algorithm for polygons.<p>The experience of working with AI agents on this project changed a lot with recent model releases, as I describe in the readme. Opus 4.8 is able to provide algorithm implementation with formal proof in one shot, whereas previous models required me to provide proof strategies in multiple steps.<p>Trust in the correctness comes entirely from the Lean checker and human review of a small specification, not from the LLM.<p>Also check out the web demo built around the verified core linked in the readme: <a href="https:&#x2F;&#x2F;schildep.github.io&#x2F;verified-polygon-intersection&#x2F;" rel="nofollow">https:&#x2F;&#x2F;schildep.github.io&#x2F;verified-polygon-intersection&#x2F;</a>. It supports multipolygons including holes, self intersections, and overlapping edges.

[IDE/Editor] Show HN: Mercek – A Desktop IDE for AWS ECS Hey HN I&#x27;ve been using ECS for a while now and found it annoying having to log into the console everytime<p>I use Lens for Kubernetes but couldnt find an equivalent for ECS so i built one!<p>The project is open source as well <a href="https:&#x2F;&#x2F;github.com&#x2F;utibeabasi6&#x2F;mercek" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;utibeabasi6&#x2F;mercek</a>

[Other] Show HN: FFmpeg WebCLI – Full FFmpeg in Browser, Offline PWA, No Uploads(WASM) Built a browser-based FFmpeg editor that runs entirely client-side via WebAssembly. Your files never leave your device -- all processing happens in a Web Worker. Works offline as an installable PWA after first load.

[Other] Anthropic's open-source framework for AI-powered vulnerability discovery

[Other] Show HN: Digger Solo – Local AI File Explorer After a lot of work I present Digger Solo 0.5.0 - the AI file explorer that respects your privacy (everything runs locally).<p>Demo video: <a href="https:&#x2F;&#x2F;vimeo.com&#x2F;1198414414" rel="nofollow">https:&#x2F;&#x2F;vimeo.com&#x2F;1198414414</a><p>New features: - LLM Chat with RAG (bring your own OpenAI compatible API key - ideally host a local model) - fresh redesign with light theme available in settings - multi-tabbed GUI: open multiple semantic maps at once - smart music player: auto-plays similar songs<p>Digger Solo offers semantic search and maps that allow you to browse your files intuively - uncovering hidden connections and near duplicates easily.<p>Happy to answer questions.

Games Between Programs: The Ruliology of Competition

OCaml Onboarding: Introduction to the Dune build system

[Other] Show HN: Hitoku Draft – Context aware local assistant Hi guys.<p>I have been working on Hitoku Draft, an open-source, voice-first AI assistant that runs entirely locally. I posted about it already, and now it has also transcription with voice editing. Looking for feedback, as I found that outside tech circles other people still do not use this tech much.<p>It&#x27;s context-aware, in the sense that it reads your screen, documents, and active app to understand what you&#x27;re working on. You can ask about PDFs, reply to emails, create calendar events, use web search, editing text, all by voice.<p>You can download a compiled version for free with the code HITOKUHN2026 <a href="https:&#x2F;&#x2F;hitoku.me&#x2F;draft&#x2F;" rel="nofollow">https:&#x2F;&#x2F;hitoku.me&#x2F;draft&#x2F;</a> (base price is 5 dollars)<p>It supports Gemma 4 and Qwen 3.5 for text generation, plus multiple STT backends (Parakeet, Qwen3-ASR).<p>Examples: - Gemma4 in action, <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=OgfI-3YjEVU" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=OgfI-3YjEVU</a> - query a pdf document, <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=ggaDhut7FnU" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=ggaDhut7FnU</a> - reply to email, <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=QFnHXMBp1gA" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=QFnHXMBp1gA</a> - and the usual voice dictation (with optional polishing)<p>I currently use it a lot with Claude Code and Logseq. Now with some friends we are also building a new cross-platform version. The goal is on the long run to have AI interactive local models serving people and professionals.

Speculative KV coding: losslessly compressing KV cache by up to ~4×

[Other] Show HN: I embedded 685M public texts in 32 minutes (on 8x A100, Rust, TensorRT) Quick note on how it works and how I&#x27;ve done my batch embedding engine IgniteMS.<p>The whole thing runs as one process using Rust, reading input, tokenizing, packing batches, keeping the queue full. TensorRT handles inference. Python is only as a wrapper.<p>I built it this way because when you use more than couple of GPUs, the GPUs stop being the problem. CPU cannot feed them fast enough. One A100 can go through batches faster than Python can tokenize and feed, so the GPU just sits there idle waiting for work. Most of my time went into optimizing this. At 8 GPUs that was basically the entire challenge.<p>On cost. I ran the big 2B messages job on a spot p4d instance (8x A100 40GB). After filtering and dedupping I got 685M raw texts. With my new engine the whole production run finishes in about half an hour. Previously I used on-demand for these jobs, now switched to spots. If AWS reclaims the box, I just rerun it. It&#x27;s roughly $7 for half-an-hour run. And at least right now spots are easier to get than on-demand.<p>Open warning: it&#x27;s batch only and NVIDIA only. You can use it both as a docker image and native. I used some optimizations for my production run. With default settings you can expect to see ~250K msg&#x2F;sec if you run the benchmark script on your p4d box. <a href="https:&#x2F;&#x2F;github.com&#x2F;Artain-AI&#x2F;ignite-ms&#x2F;blob&#x2F;main&#x2F;BENCHMARKING.md" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;Artain-AI&#x2F;ignite-ms&#x2F;blob&#x2F;main&#x2F;BENCHMARKIN...</a><p>v1.1.0 added TensorRT 11 and 60 models, 23 tested on 1x and 4x A100.<p>Happy to share details.

[Other] KVarN: Native vLLM backend for KV-cache quantization by Huawei

[DevOps] Show HN: Boxes.dev: ditch localhost; run Claude Code and Codex in the cloud Hi HN, we’re Nick and Drew, and we’re building boxes.dev – the first cloud-only agentic dev environment (ADE) that gives every Codex and Claude Code agent its own cloud computer.<p>We’re two engineers who previously built Gem (co-founder&#x2F;CTO and first hire), and we spent the last year coding almost exclusively using Codex and Claude Code. It’s been a huge change to how we code, and it’s been exhilarating seeing the models keep getting better – but we eventually realized that developing on localhost was holding us back:<p>- Git worktrees are clunky to set up and use for parallelizing work - It’s 2026, but somehow everyone is still walking around with laptops cracked open or SSHing into mac minis in their garage so their agents don’t stop working. - Mobile is treated like an afterthought even though coding is just texting now We started hitting resource constraints when multiple parallel agents test their own work by running the full app locally. - We tried different products, but couldn’t find any that solved all of our pain points – so we pivoted and decided to just build the ADE we wanted for ourselves.<p>Boxes.dev is a desktop and mobile app that lets you run Claude Code, Codex (using your subscription!), and the full dev environment for whatever you’re building, all on remote compute. It’s similar to Conductor or the Codex desktop app, except everything is in the cloud.<p>We use coding agents to scan your local dev setup and port it to the cloud. Then every Claude Code&#x2F;Codex thread starts from a snapshot of the full setup, with its own filesystem and compute. No more git worktrees, no more cracked-open laptops, and your coding agents can actually test their work end-to-end because they can run your full app in isolation.<p>We’ve mirrored the Claude Code and Codex UX to feel natural to power users, and also have a fully-featured mobile app (no handoffs or remote control), plus scheduled automations and a Slack integration.<p>We’re obviously biased, but we’ve been building boxes.dev with boxes.dev for months and it’s honestly been a gamechanger. It’s hard to go back once you realize how much localhost has been limiting you; based on early feedback from beta testers, we’re increasingly sure that cloud is the future of agentic coding.<p>We’d love for you to experience it yourselves! Would appreciate any feedback – and happy to answer any questions on this thread.

The Secret Life of Circuits with lcamtuf / Michał Zalewski (Audio Interview)

Europe's largest Copper Age tomb: children's bones show ancient health crisis

[Other] Show HN: Hydron – Hardware-aware coding agent Hi HN, this is Prashant from H2Loop. Embedded engineers that we work with were annoyed that generic AI tools hallucinated register addresses, generated code for peripherals that don&#x27;t exist on the chip and mixed up timer quirks between similar platforms like STM32F4 and F7. The code looks clean but it just won&#x27;t boot. This made them go back to the datasheet every time. So we built Hydron, an AI tool that writes datasheet-grounded code for your hardware.<p>Demo: Hydron setting up sleep-mode CPU logging for an onboard temp sensor on an STM32U385 - <a href="https:&#x2F;&#x2F;boot.hydron.sh&#x2F;zzzDemo" rel="nofollow">https:&#x2F;&#x2F;boot.hydron.sh&#x2F;zzzDemo</a><p>First, we&#x27;ve pre-indexed 580+ platforms and peripherals. Most of what you&#x27;d use in a robotics, UAV, or IoT build: common dev platforms like STM32, ESP32, RP2040, AM6 families, plus the IMUs, GNSS modules, motor drivers, and baros that ship around them. Ask about a peripheral, the answer comes from our KG and the actual datasheet.<p>Second, you can bring your own context and share it with your team. Hydron indexes PDFs up to 5000 pages, plus a whole host of various file types and even ZIPs of full C&#x2F;C++&#x2F;Python codebases up to 250MB. One engineer indexes the HW, BSPs, and datasheet pack once. Anyone else can reference it from their own Hydron agent.<p>Third, HW-SW development happens in your editor or terminal. Agentic serial monitor is live today. GDB integration, and an AI log reader land in the next two weeks.<p>Up next, we&#x27;re focused on closing the hardware-software loop. We&#x27;re building more HIL debugging capabilities, deeper target awareness, and support for additional platforms. If you work on embedded SW we&#x27;d love your feedback on where today&#x27;s tools fall short and what you&#x27;d like to see next.<p>Install: VS Code extension - <a href="https:&#x2F;&#x2F;marketplace.visualstudio.com&#x2F;items?itemName=H2Loop.hydron" rel="nofollow">https:&#x2F;&#x2F;marketplace.visualstudio.com&#x2F;items?itemName=H2Loop.h...</a> CLI mac&#x2F;linux -&gt; curl -fsSL <a href="https:&#x2F;&#x2F;get.hydron.sh&#x2F;cli&#x2F;install.sh" rel="nofollow">https:&#x2F;&#x2F;get.hydron.sh&#x2F;cli&#x2F;install.sh</a> | bash. CLI windows -&gt; irm <a href="https:&#x2F;&#x2F;get.hydron.sh&#x2F;cli&#x2F;install.ps1" rel="nofollow">https:&#x2F;&#x2F;get.hydron.sh&#x2F;cli&#x2F;install.ps1</a> | iex<p>More at <a href="http:&#x2F;&#x2F;boot.hydron.sh&#x2F;HN" rel="nofollow">http:&#x2F;&#x2F;boot.hydron.sh&#x2F;HN</a>. 200 free credits, 50% off on paid plans, one-step signup. Me and u&#x2F;ajithhyd will be in the thread all day.

[CLI Tool] Show HN: Cost.dev (YC W21) – making agents cost-aware and cheaper to call We launched Infracost on HN five years ago (<a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=26064588">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=26064588</a>) where our CLI generated cost estimates for infra-as-code, e.g. &quot;this Terraform PR adds $400&#x2F;mo&quot;. The idea was to shift cloud costs (FinOps) left, so engineers get visibility of costs before deployment and make better decisions.<p>Earlier this year we started seeing agent traffic in our logs and it looked like coding agents were calling our CLI. But that CLI wasn&#x27;t designed with coding agents in mind. We went down a philosophical rabbit hole to see if a CLI is even needed anymore given that Claude, Copilot et al. already follow best practices. Ultimately we decided to create a new CLI from the ground up with coding agents in mind for two reasons:<p>1. We optimized the CLI for agent callers and cut Claude&#x27;s output token usage by up to 79% and API cost by up to 67% versus a bare-Claude baseline. We wrote a blog documenting our lessons on optimizing user token usage when designing a CLI, e.g. using predicate flags so the agent doesn&#x27;t compose jq | python | wc pipelines, output format that strips JSON&#x27;s redundant field names. The blog is here: <a href="https:&#x2F;&#x2F;www.infracost.io&#x2F;resources&#x2F;blog&#x2F;we-cut-claude-s-token-usage-79-by-redesigning-our-cli-for-agents">https:&#x2F;&#x2F;www.infracost.io&#x2F;resources&#x2F;blog&#x2F;we-cut-claude-s-toke...</a><p>2. With cloud costs, precision matters. Telling a coding agent &quot;make this Terraform cost-optimized&quot; can be expensive and lossy. You burn tokens loading code and policy context into every conversation. Your agent could make up a price and you wouldn&#x27;t know because it&#x27;s difficult to verify that across the ~10M price points that AWS, Azure and Google have. The CLI runs static analysis on the code, uses the latest prices from cloud vendors, and passes that context to the coding agent.<p>So that&#x27;s what we&#x27;re launching today - Cost.dev: <a href="https:&#x2F;&#x2F;cost.dev&#x2F;" rel="nofollow">https:&#x2F;&#x2F;cost.dev&#x2F;</a>.<p>- It runs locally. Your code never leaves your machine, you get a fast feedback loop, and you&#x27;re not burning API calls per character when you want to fetch prices.<p>- The CLI does the deterministic work. Fetching price points, scanning the code, validating fixes. The coding agent does the natural-language part. You don&#x27;t have to trust the LLM to remember the rules, and can verify it called the right CLI command.<p>- It provides a consistent rule layer across every tool you use. Get cost estimates in your IDE and your coding agent with a single install. We support Claude Code, GitHub Copilot, Cursor, Windsurf, OpenAI Codex, Gemini CLI, as well as IDEs like VS Code and JetBrains<p>Before we keep building more in that direction, I want to sanity-check with HN: is &quot;agents writing IaC in prod&quot; actually a thing yet, or am I betting on a future that&#x27;s still a year out? I know software developers are using coding agents heavily, but are platform&#x2F;infra folks doing that for prod too? Also, if you have any feedback on Cost.dev, I&#x27;d love to hear it!

Summer of '85: DOSBOS is rejected by ANALOG Computing