snaps - Moozonian Search

github.com• Mar 6, 2026• 1 min read

Show HN: Moongate – Ultima Online server emulator in .NET 10 with Lua scriptingI've been building a modern Ultima Online server emulator from scratch. It's not feature-complete (no combat, no skills yet), but the foundation is solid and I wanted to share it early.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/RunUO)Tech stack: .NET 10, NativeAOT, NLua, MessagePack, DryIoc, KestrelWhat's missing: Combat, skills, weather integration, NPC AI. This is still early — the

github.com• Mar 6, 2026• 1 min read

Show HN: A library to turn any React Native View into a video or an imageIntroducing React Native View Recorder!The ultimate library to record any native View in your app and export it as a video or an image. No server required, MIT Licensed.- Record or snapshot any arbitrary View in React Native, whole app dashboard or only part of it - Supports recording Skia Canvas - You have frame-by-frame control over the video you're recording - Add your audio file or mix in the audio frame by frame - Fully native, no FFmpeg binaries - Control FPS, bitrate - Support for H.264, HEVC, and transparencyStar it on GitHub: https://github.com/Rednegniw/react-native-view-recorderCheck out the docs: https://react-native-view-recorder.awingender.com/

bing.com• Mar 5, 2026• 1 min read

Bears moving on without DJ Moore, but memories of his biggest moments will lastBears fans will never forget DJ Moore's winning touchdown catch against the Packers in Week 16. Michael Reaves / Getty Images For those who want the snapshot of DJ Moore reaching for one of the ...

news.ycombinator.com• Mar 5, 2026• 1 min read

Fly.io deleted my apps and DBs in an unrelated organization without warningHi HN,I’m experiencing a destructive and seemingly erratic behavior from Fly.io's automated systems that resulted in a total loss of data across multiple organizations.I have been using Fly.io since July 2023 with two distinct organizations:Org A: Used for some Selenium-based automation experiments in late 2025 and other unrelated private projects through the years.Org B: A long-standing production environment with a Postgres database, active since 2023 with a consistent billing history.The Incident Recently, I discovered that everything across all my organizations has been nuked. Apps, databases, and even the volume snapshots (backups) have vanished. The dashboard for my database volume now simply says "deleted 16 days ago."The contradiction in their system is staggering:In both organizations, I see a semi-transparent banner: "Your account was flagged by our fraud protection system. Learn more" (the "Learn more" is not even a link).However, the "Billing" page for Org B still claims th

news.ycombinator.com• Feb 28, 2026• 1 min read

Show HN: PyTorch/FEniCSx pipeline for elastocaloric metamaterial optimizationHi HN,I built an open-source Deep Reinforcement Learning (DRL) environment to explore topology optimization for elastocaloric cooling (solid-state HVAC).The physical problem: The HVAC industry is trying to replace toxic HFC refrigerants with elastocaloric Shape Memory Alloys like Nitinol. You cool a room by cyclically compressing a porous metal lattice. The paradox is that you need high porosity for air to flow through the metal, but holes create stress concentrators. If the metal is too thin, it snaps after a few thousand cycles from mechanical fatigue. Standard density-based topology optimization (like SIMP) usually outputs jagged, unprintable "grayscale" meshes.What I built: I wanted to bypass SIMP and enforce manufacturing reality. I wrote an OpenAI Gymnasium wrapper that bridges a PyTorch RL agent with a native CAD-to-Physics pipeline: 1. The PyTorch agent outputs continuous geometric parameters for BCC/Gyroid lattices. 2. The environment checks the resulting wall thickness. If it

unfudged.io• Feb 26, 2026• 1 min read

Show HN: Unfucked - version all changes (by any tool) - local-first/source availI built unf after I pasted a prompt into the wrong agent terminal and it overwrote hours of hand-edits across a handful of files. Git couldn't help because I hadn't finished/committed my in progress work. I wanted something that recorded every save automatically so I could rewind to any point in time. I wanted to make it difficult for an agent to permanently screw anything up, even with an errant rm -rfunf is a background daemon that watches directories you choose (via CLI) and snapshots every text file on save. It stores file contents in an object store, tracks metadata in SQLite, and gives you a CLI to query and restore any version. The install includes a UI, as well to explore the history through time.The tool skips binaries and respects `.gitignore` if one exists. The interface borrows from git so it should feel familiar: unf log, unf diff, unf restore.I say "UN-EF" vs U.N.F, but that's for y'all to decide: I started by calling the project Unfucked and got unfucked.ai, which if you

github.com• Feb 26, 2026• 1 min read

Show HN: MongoClaw, A mutation runtime for MongoDB with write-time agent safetyI built MongoClaw because every AI enrichment pipeline I've seen hits the same concurrency bug nobody talks about.Your agent reads a document at t0. Inference takes 2 seconds. At t1, another process updates that document. At t2, the agent writes back — silently overwriting live data with output generated from a stale snapshot. Clean 200 OK. No error.That's not a prompt problem. It's a write-safety problem.MongoClaw solves it by capturing version + content hash at dispatch time and issuing a conditional write that only succeeds if the source record still matches. Stale payload? Write suppressed, reason classified, execution record persisted.It also handles:* Idempotent replay protection across all write strategies * Loop detection via agent-origin metadata * In-band policy evaluation (enrich/block/tag) after inference and before mutationAgents are declared in YAML. The runtime handles change stream ingestion, Redis-backed queuing, async execution, schema validation, and auditable writeb

news.ycombinator.com• Feb 24, 2026• 1 min read

Show HN: Pilo – open-source agentic web automation engine by MozillaHello HN,We are the team behind Tabstack (https://tabstack.ai) - part of Mozilla. We just open sourced Pilo (pronounce PIE-low), the core engine that powers our automation platform. You can check it out on Github at https://github.com/mozilla/pilo.Pilo is an agentic web automation library. Instead of writing rigid scripts with CSS selectors, you give it a natural language goal (e.g., "Find the best pizza in Seattle and extract the ratings") and it autonomously navigates the browser to achieve it.We built this because we were struggling to make reliable agents for our own /automate endpoint. Existing tools were either too brittle (breaking on minor DOM changes) or too heavy (feeding raw HTML to LLMs, blowing up context windows).Here is how Pilo solves those problems:- Accessibility Tree over HTML: Instead of parsing raw HTML "soup," Pilo captures the browser's accessibility tree (via Playwright's _snapshotForAI). This gives the LLM a semantic, stable view of the page (buttons, links, in

news.ycombinator.com• Feb 23, 2026• 1 min read

Managed OpenClaw hosting, 60-second provisioningI registered https://clawhosters.com 18 days ago. Wrote the first line of code that same day. The whole thing is live now with paying customers, so I figured it was time to share it here.The problem: deploying OpenClaw (the open-source AI agent framework) is annoying. Docker setup, Nginx config, SSL, firewall rules, Playwright browser installation for web tools, messenger bridge configuration. If you know what you're doing, it probably takes an hour. I got it down to 30-60 seconds.*The provisioning trick: prewarmed VPS*Snapshot-based provisioning alone gets you to about 3 minutes. Hetzner needs ~40s to create the VPS from a snapshot alone.The thing that actually makes it fast is prewarming. I keep 1-2 idle VPS per tier sitting ready. They're already booted from the snapshot, Docker is running, everything is warm. When a customer clicks "Create Instance," I claim a prewarmed server, SSH in, upload configs, start containers. 30-60 seconds, done.A background job monitors the pool and repl

github.com• Feb 22, 2026• 1 min read

Show HN: Madari: CLI tool to install, sync and manage local MCP serversI built stewreads-mcp (https://github.com/ankitvg/stewreads) to turn useful AI chats into ebooks and send them to Kindle. Every time I wanted to share it I had to explain how to find the binary, edit claude_desktop_config.json, get the JSON update exactly right, and restart Claude Desktop. Too much friction.So I built madari. Now, stewreads-mcp can be installed and setup with Claude Desktop in just one command:madari install stewreads-mcpThat one command runs uv tool install, resolves the binary path, registers the server, and syncs it to Claude Desktop.Madari also includes: lifecycle commands (add, remove, enable, disable), sync to push changes to the client config, diagnostics (doctor, status) to monitor MCP health, and snapshot export/import for moving your setup across machines. Claude Desktop is the supported client; more client support coming next.Would love feedback from folks building MCP servers who's hit the same problem.Tame your MCP servers. https://github.com/ankitvg/madar

vistaley.com• Feb 17, 2026• 1 min read

Show HN: We built a free VC platform that shares data between GPS and foundersHey HN,We built Vistaley — a two-sided platform for VC fund management. The GP-facing side (VentureLens) handles fund operations: deal pipeline, portfolio tracking, fund accounting, LP reporting. The founder-facing side (Harbour) provides free FP&A tools: financial dashboards, KPI tracking, burn rate analysis.When a portfolio company enters their financials in Harbour, the data is immediately available in the GP's VentureLens dashboard — no CSV exports, no quarterly spreadsheets, no API integrations.Interesting challenges we solved:- Multi-currency — 60+ currencies with local display toggle. Every financial metric can be viewed in fund currency or portfolio company's local currency. Exchange rates stored per-snapshot, not globally, to preserve historical accuracy.- Multi-jurisdiction accounting — 12 jurisdictions (including Kazakhstan, Pakistan, Bangladesh) with different tax frameworks, compliance requirements, and regulatory reporting standards. One accounting standard per fund, enfo

github.com• Feb 16, 2026• 1 min read

Show HN: Jsiphon – Streaming JSON parser with delta tracking and ambiguity treesHi HN, I built Jsiphon to solve a common frustration with LLM streaming: you ask for structured JSON output, but can't use any of it until the entire stream finishes.If you've used JSON mode (OpenAI, Anthropic, etc.), you've hit this — you want {"answer": "...", "sources": [...]}, but JSON.parse() fails on every incomplete chunk.LLM responses are inherently append-only (tokens arrive left to right, never go back), so Jsiphon leans into that with three ideas:1) Append-only parsing — Feed in {"msg": "Hel and get {msg: "Hel"} immediately. Values are only extended, never removed or mutated.2) Delta tracking — Each snapshot contains only what's new. For a chat bubble, just append delta.content to the DOM — when the LLM produces next chunk "lo, World!", we immediately get {msg: "lo, World!"}. No need to repeat partial JSON parsing or full tree rerendering.3) Ambiguity tree — A tree that mirrors the shape of your data and tracks which subtrees are finalized at every depth. For example, if you

news.ycombinator.com• Jan 28, 2026• 1 min read

Show HN: VNOL – The Vendor-Neutral Cognitive OS Layer for Agent PortabilityHey HN,I’ve been obsessed with the idea that while LLMs are becoming commodities, Agent State is becoming the new vendor lock-in.Currently, if you build an agent in LangChain, its "mind" (history, intent-graph, learned synapses) is trapped there. If you want to move that same cognitive session from GPT-4 to Claude 3.5, or from one framework to another, you’ve basically got to start from scratch or deal with massive context loss.I built VNOL (Vendor-Neutral Orchestration Layer) to solve this at the kernel level.What it is: VNOL is a "Cognitive OS" layer. It treats an agentic session like a process that can be snapshotted, serialized (VNOL v2.0 YAML spec), and rehydrated on any model or framework.Key Technical Hooks:Pre-Disk Intent Locking: We intercept agent actions (file writes, API calls) before they hit the system, with a 12ms latency budget (using Azhwar, a hybrid Regex/AST engine). Mind Portability: Decouples the reasoning skeleton from the model. You can "pause" a session on OpenA

sentienceapi.com• Jan 28, 2026• 1 min read

A verification layer for browser agents: Amazon case studyA common approach to automating Amazon shopping or similar complex websites is to reach for large cloud models (often vision-capable). I wanted to test a contradiction: can a ~3B parameter local LLM model complete the flow using only structural page data (DOM) plus deterministic assertions?This post summarizes four runs of the same task (search → first product → add to cart → checkout on Amazon). The key comparison is Demo 0 (cloud baseline) vs Demo 3 (local autonomy); Demos 1–2 are intermediate controls.More technical detail (architecture, code excerpts, additional log snippets):https://www.sentienceapi.com/blog/verification-layer-amazon-...Demo 0 vs Demo 3:Demo 0 (cloud, GLM‑4.6 + structured snapshots) success: 1/1 run tokens: 19,956 (~43% reduction vs ~35k estimate) time: ~60,000ms cost: cloud API (varies) vision: not requiredDemo 3 (local, DeepSeek R1 planner + Qwen ~3B executor) success: 7/7 steps (re-run) tokens: 11,114 time: 405,740ms cost: $0.00 incremental (local inference) vi

theideon.com• Jan 26, 2026• 1 min read

Show HN: Ideon – An open source, infinite canvas for your project's segmentationThe problem: I want to use my tools. You want to use yours. The contractor we hired uses yet another set of tools. This causes unnecessary friction for all participants.Ideon is a self-hosted visual workspace designed to bridge this gap. It doesn't replace your existing stack (GitHub, Figma, Notion, etc.) but provides a shared context where all these pieces live together on an infinite canvas.We built this because projects often die from fragmentation—code is in one place, decisions in chat logs, and visuals in design tools. Ideon aims to keep the project "mentally navigable" for everyone involved.Key features: - Visual Blocks: Organize Repositories, Notes, Links, Files, and People spatially. - State History: Track how decisions evolved with workspace snapshots. - Multiplayer: Real-time collaboration. - Self-hosted: Docker-based and AGPLv3 licensed.Tech stack: Next.js, PostgreSQL, Docker.Would love to hear your feedback on the approach!

github.com• Jan 15, 2026• 1 min read

Show HN: I'm building an open-source AI agent runtime using Firecracker microVMsHello Hacker News! I'm Mark. I'm building Moru, an open-source runtime for AI agents that runs each session in an isolated Firecracker microVM. It started as a fork of E2B, and most of the low-level Firecracker runtime is still from upstream.It lets you run agent harnesses like Claude Code or Codex in the cloud, giving each session its own isolated microVM with filesystem and shell access.The repo is: https://github.com/moru-ai/moruEach VM is a snapshot of a Docker build. You define a Dockerfile, CPU, memory limits, and Moru runs the build inside a Firecracker VM, then pauses and saves the exact state: CPU, dirty memory pages, and changed filesystem blocks.When you spawn a new VM, it resumes from that template snapshot. Memory snapshot is lazy-loaded via userfaultfd, which helps sandboxes start within a second.Each VM runs on Firecracker with KVM isolation and a dedicated kernel. Network uses namespaces for isolation and iptables for access control.From outside, you talk to the VM thro

diffswarm.dev• Jan 6, 2026• 1 min read

Show HN: Diffswarm – GitHub-style review and workflow for any unified diffPipe the output of a `diff` command to Diffswarm to get back a persistent URL that contains the diff with workflow, comments, search, etc. Basically some of the helpful features that Github offers on PR diffs, just available for any arbitrary diff.I wrote this tool to use at work for documenting snapshot diff triaging, and also used it to help with the same thing on a couple Ty PRs. Example: https://diffswarm.dev/d-01k2gknwyq82f6x17zqf3apjxc

github.com• Dec 17, 2025• 1 min read

Show HN: I built a deterministic vector DB kernel in Rust using fixed-point mathWhile building a vector database in Python (valori: https://pypi.org/project/valori/), I ran into a bug that took longer than I want to admit.The same embeddings, indexed on macOS and Windows, would sometimes produce slightly different nearest-neighbor results. Nothing dramatic but enough to break tests and make snapshots non-reproducible.At first I assumed it was my code. Then I realized the problem wasn’t my logic at all it was "floating-point math".Different CPUs, different compilers, FMA instructions, and subtle differences in IEEE behavior meant that “the same” computation wasn’t actually the same everywhere.That sent me down a rabbit hole.The idea:Instead of trying to fight floating-point nondeterminism, I asked a simpler question:What if the core of a vector system never used floats at all?So I built a small kernel in Rust that uses fixed-point arithmetic (Q16.16) for all vector math. No floats in the core. No randomness. No architecture-dependent behavior.Same inputs → same sta

bing.com• Dec 16, 2025• 1 min read

I Tested Makita’s New Modular Toolbox System—And It Changed How I Store ToolsThe MakTrak system snaps together like Lego bricks and rolls easily from garage to truck, but its low-profile design comes with trade-offs. Recently, I tested out the Makita Tools MakTrak modular ...

bing.com• Nov 25, 2025• 1 min read

The 9 Best Digital Picture Frames That Can Replace Your Traditional Wall HangingVacation snaps, family pictures, pet portraits—they’re probably all trapped on your phone’s camera roll, and you’re not going to leave your phone propped on a shelf to display those photos. But with a ...