🛠️ All DevTools

Entomologists use a particle accelerator to image ants at scale

Show HN: Moongate – Ultima Online server emulator in .NET 10 with Lua scripting I&#x27;ve been building a modern Ultima Online server emulator from scratch. It&#x27;s not feature-complete (no combat, no skills yet), but the foundation is solid and I wanted to share it early.<p>What it does today: - Full packet layer for the classic UO client (login, movement, items, mobiles) - Lua scripting for item behaviors (double-click a potion, open a door — all defined in Lua, no C# recompile) - Spatial world partitioned into sectors with delta sync (only sends packets for new sectors when crossing boundaries) - Snapshot-based persistence with MessagePack - Source generators for automatic DI wiring, packet handler registration, and Lua module exposure - NativeAOT support — the server compiles to a single native binary - Embedded HTTP admin API + React management UI - Auto-generated doors from map statics (same algorithm as ModernUO&#x2F;RunUO)<p>Tech stack: .NET 10, NativeAOT, NLua, MessagePack, DryIoc, Kestrel<p>What&#x27;s missing: Combat, skills, weather integration, NPC AI. This is still early — the focus so far has been on getting the architecture right so adding those systems doesn&#x27;t require rewiring everything.<p>Why not just use ModernUO&#x2F;RunUO? Those are mature and battle-tested. I started this because I wanted to rethink the architecture from scratch: strict network&#x2F;domain separation, event-driven game loop, no inheritance-heavy item hierarchies, and Lua for rapid iteration on game logic without recompiling.<p>GitHub: <a href="https:&#x2F;&#x2F;github.com&#x2F;moongate-community&#x2F;moongatev2" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;moongate-community&#x2F;moongatev2</a>

I Dropped Our Production Database and Now Pay 10% More for AWS

Select context for coding agents directly from your website

基于 Claude Code 的长篇网文辅助创作系统，解决 AI 写作中的「遗忘」和「幻觉」问题，支持 200 万字量级 连载创作。

CyberStrikeAI is an AI-native security testing platform built in Go. It integrates 100+ security tools, an intelligent orchestration engine, role-based testing with predefined security roles, a skills system with specialized testing skills, and comprehensive lifecycle management capabilities.

Agent framework and applications built upon Qwen>=3.0, featuring Function Calling, MCP, Code Interpreter, RAG, Chrome extension, etc.

Hardening Firefox with Anthropic's Red Team The bugs are the ones that say &quot;using Claude from Anthropic&quot; here: <a href="https:&#x2F;&#x2F;www.mozilla.org&#x2F;en-US&#x2F;security&#x2F;advisories&#x2F;mfsa2026-13&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.mozilla.org&#x2F;en-US&#x2F;security&#x2F;advisories&#x2F;mfsa2026-1...</a><p><a href="https:&#x2F;&#x2F;blog.mozilla.org&#x2F;en&#x2F;firefox&#x2F;hardening-firefox-anthropic-red-team&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blog.mozilla.org&#x2F;en&#x2F;firefox&#x2F;hardening-firefox-anthro...</a><p><a href="https:&#x2F;&#x2F;www.wsj.com&#x2F;tech&#x2F;ai&#x2F;send-us-more-anthropics-claude-sniffs-out-bevy-of-bugs-c6822075" rel="nofollow">https:&#x2F;&#x2F;www.wsj.com&#x2F;tech&#x2F;ai&#x2F;send-us-more-anthropics-claude-s...</a>

Show HN: Moltty – Organized, Persistent AI Coding Sessions

[Other] Show HN: 1v1 coding game that LLMs struggle with This is a game I wish I had as a kid learning programming. The concept of it is fairly similar to other coding games like Screeps, but instead of a complex world with intricate mechanics, Yare is a lot more minimal and approachable with quick 1v1 &lt;3 min matches.<p>It&#x27;s purely a passion project with no monetization aspirations. And it&#x27;s open source: <a href="https:&#x2F;&#x2F;github.com&#x2F;riesvile&#x2F;yare" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;riesvile&#x2F;yare</a><p>The first version &#x27;launched&#x27; several years ago and I got some good feedback here: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=27365961">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=27365961</a> that I iterated on.<p>The latest overhaul is a result of simplifying everything while still keeping the skill ceiling high. And at least the LLMs seem to struggle with this challenge for now (I run a small tournament between major models - results and details here: <a href="https:&#x2F;&#x2F;yare.io&#x2F;ai-arena" rel="nofollow">https:&#x2F;&#x2F;yare.io&#x2F;ai-arena</a><p>I&#x27;d love to hear your thoughts

Show HN: Swarm – Program a colony of 200 ants using a custom assembly language We built an ant colony simulation as an internal hiring challenge at Moment and decided to open it up publicly.<p>You write a program in a custom assembly-like (we call it ant-ssembly) instruction set that controls 200 ants. Each ant can sense nearby cells (food, pheromones, home, other ants) but has no global view. The only coordination mechanism is pheromone trails, which ants can emit and sense them, but that&#x27;s it. Your program runs identically on every ant.<p>The goal is to collect the highest percentage of food across a set of maps. Different map layouts (clustered food, scattered, obstacles) reward very different strategies. The leaderboard is live.<p>Grand prize is a trip to Maui for two paid for by Moment. Challenge closes March 12.<p>Curious what strategies people discover. We&#x27;ve seen some surprisingly clever emergent behavior internally.

Show HN: Moji – A read-it-later app with self-organizing smart collections I built Moji because I was drowning in saved articles. Every read-it-later app I tried became a graveyard of unread links — no structure, no way to surface the right article at the right time. Moji is a native iOS read-it-later app that saves articles for offline reading and organizes them automatically using smart collections. The name &quot;Moji&quot; comes from 墨迹 (mòjì) in Chinese — it literally means &quot;ink traces,&quot; but colloquially it means being slow or dawdling. Felt fitting for an app that lets you save things to read later — no rush.<p>Smart Collections — the core idea<p>Instead of manually tagging or filing articles, you define criteria and Moji continuously filters your library for you. Criteria combine with AND logic between types and OR logic within a type, so you can build surprisingly precise filters:<p>- Domain: arxiv.org, paperswithcode.com + Saved: This Week → Fresh ML papers<p>- Keywords: &quot;SwiftUI&quot;, &quot;Combine&quot; + Unread → Your iOS learning queue<p>- Reading Time: &gt; 15 min + Unread → Weekend deep dives<p>- Domain: news.ycombinator.com + Saved: Last 7 days → This week&#x27;s HN saves<p>- Language: zh + Reading Time: &lt; 5 min → Quick Chinese reads for your commute<p>Four system collections come built in — Unread, Quick Reads (&lt;5 min), Deep Dive (&gt;10 min), and This Week — so it&#x27;s useful out of the box. Pin your favorites to the filter bar for one-tap access.<p>Other features<p>- Native SwiftUI reader — Articles render as native SwiftUI views, not a WebView. This means real offline reading, smooth scrolling, and proper typography controls (font size, serif&#x2F;sans-serif, line spacing).<p>- On-device AI summaries — One-sentence TL;DRs powered by Apple Intelligence. Runs entirely on-device, no cloud calls. Supports 10+ languages.<p>- Full-text search — Search across titles and content with context snippets that jump you straight to the match in the article.<p>- Reading position memory — Remembers exactly where you left off, down to the block and scroll offset.<p>- Image viewer — Pinch-to-zoom, double-tap, pan, alt-text display.<p>- PDF export — Save any article as a styled PDF.<p>- Share extension — Save from Safari in two taps.<p>- Language-aware reading time — Calculates differently for CJK (260 WPM) vs. English (200 WPM) vs. Arabic&#x2F;Hebrew (150 WPM).<p>- iCloud sync — Optional CloudKit sync across devices.<p>- Privacy-first — All processing happens on-device. No analytics, no tracking.<p>Technical details for the curious<p>Built with Swift 6.2, SwiftData, structured concurrency, and Mozilla&#x27;s Readability.js for content extraction. The HTML parser converts articles into typed ContentBlock values that SwiftUI renders natively. A three-phase background pipeline handles extraction, quality re-extraction, and summary generation.<p>Pricing<p>Start with a 2-week free trial — all features unlocked, no restrictions. After that, a one-time Pro purchase ($9.99 in US, price may vary in other countries) is required to save new articles. No subscription. You never lose access to your existing library, reading features, or smart collections — the gate is only on adding new articles.<p>I&#x27;d love feedback — especially on the smart collection criteria. What filters would make this more useful for your workflow?<p>---<p>App Store Link: <a href="https:&#x2F;&#x2F;apps.apple.com&#x2F;us&#x2F;app&#x2F;moji-reader&#x2F;id6758530352">https:&#x2F;&#x2F;apps.apple.com&#x2F;us&#x2F;app&#x2F;moji-reader&#x2F;id6758530352</a>

Labor market impacts of AI: A new measure and early evidence

Ethiopia gets $350M World Bank financing for its digital ID project (2024)

Proton Mail Helped FBI Unmask Anonymous 'Stop Cop City' Protester

Show HN: Docker pulls more than it needs to - and how we can fix it Hi all!<p>I&#x27;ve built a small tool to visualize how inefficient `docker pull` is, in preparation for standing up a new Docker registry + transport. It&#x27;s bugged me for a while that updating one dependency with Docker drags along many other changes. It&#x27;s a huge problem with Docker+robotics. With dozens or hundreds of dependencies, there&#x27;s no &quot;right&quot; way to organize the layers that doesn&#x27;t end up invalidating a bunch of layers on a single dependency update - and this is ignoring things like compiled code, embedded ML weights, etc. Even worse, many robotics deployments are on terrible internet, either due to being out in the boonies or due to customer shenanagins. I&#x27;ve been up at 4AM before supporting a field tech who needs to pull 100MB of mostly unchanged Docker layers to 8 robots on a 1Mbps connnection. (and I don&#x27;t think that robotics is the only industry that runs into this, either - see the ollama example, that&#x27;s a painful pull)<p>What if Docker were smarter and knew about the files were already on disk? How many copies of `python3.10` do I have floating around `&#x2F;var&#x2F;lib&#x2F;docker`. For that matter, how many copies of it does DockerHub have? A registry that could address and deduplicate at the file level rather than just the layer level is surely cheaper to run.<p>This tool:<p>- Given two docker images, one you have and one you are pulling, finds how much data docker pull would use, as well as how much data is _actually_ required to pull<p>- Shows an estiimate for how much time you will save on various levels of cruddy internet<p>- There&#x27;s a bunch of examples given of situations where more intelligent pulls would help, but the two image names are free text, feel free to write your own values there and try it out (one at a time though, there&#x27;s a work queue to analyze new image pairs)<p>The one thing I wish it had but haven&#x27;t gotten around to fitting in the UI somehow is a visualization of the files that _didn&#x27;t_ change but are getting pulled anyhow.<p>It was written entirely in Claude Code, which is a new experience for me. I don&#x27;t know nextjs at all, I don&#x27;t generally write frontends. I could have written the backend maybe a little slower than Claude, but the frontend would have taken me 4x as long and wouldn&#x27;t have been as pretty. It helped that I knew what I wanted on the backend, I think.<p>The registry&#x2F;transport&#x2F;snapshotter(?) I&#x27;m building will allow both sharing files across docker layers on your local machine well as in the registry. There&#x27;s a bit of prior art with this, but only on the client side. The eStargz format allows splitting apart the metadata for a filesystem and the contents, while still remaining OCI compliant - but it does lazy pulls of the contents, and has no deduplication. I think it could easily compete with other image providers both on cost (due to using less storage and bandwidth...everywhere) as well as speed.<p>If you&#x27;d be interested, please reach out.

GLiNER2: Unified Schema-Based Information Extraction

[Other] SQG (SQL to Code Generator) v0.10: Java Streams and List Type Support

Show HN: Vet – Prevent coding agents from making mistakes

Seventeen Years of Coding and Starting Over