Embed - 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

news.ycombinator.com• Mar 6, 2026• 1 min read

Malicious NPM package pino-SDK-v2 exfiltrates .env secrets to DiscordWe just analyzed a fresh supply chain attack on npm that's pretty well-executed.Package: pino-sdk-v2 Target: Impersonates pino (one of the most popular Node.js loggers, ~20M weekly downloads)Reported to OSV too- https://osv.dev/vulnerability/MAL-2026-1259What makes this one interesting: The attacker copied the entire pino source tree, kept the real author's name (Matteo Collina) in package.json, mirrored the README, docs, repository URL so everything looks legitimate on the npm page.The only changes: - Renamed package to pino-sdk-v2 - Injected obfuscated code into lib/tools.js (300+ line file) - No install hooks whatsoeverThe payload: Scans for .env, .env.local, .env.production, .env.development, .env.example files, extracts anything matching PRIVATE_KEY, SECRET_KEY, API_KEY, ACCESS_KEY, SECRET, or just KEY=, then POSTs it all to a Discord webhook as a formatted embed.The malicious function is literally named log(). In a logging library. That's some next-level camouflage.Why most scann

github.com• Mar 5, 2026• 1 min read

Show HN: Msplat – 3D Gaussian Splatting training in ~90s on M4 Max, native MetalHey HN, I built msplat because I wanted to train 3DGS scenes on my Mac without pulling in torch. Most ports I came across go through autograd and hence come with ~2GB of framework overhead, which felt overkill for a pipeline that's just a few dozen GPU kernels + an optimizer.So I wrote the whole training pipeline from scratch as Metal shaders: projection, tile-based rasterization, SSIM loss, backward pass, Adam, and densification. Everything runs on the GPUmsplat trains 7k iterations of full-resolution Mip-NeRF 360 scenes in ~90s on my M4 Max. In the README I compare against gsplat's published numbers, which were measured on a TITAN RTX. Ofc these are different hardware classes, so take the wall-time comparisons with a grain of saltPython bindings are on PyPI (pip install msplat), and there are Swift bindings if you want to embed this in a native app. Happy to answer questions about any of the internalsRepo: https://github.com/rayanht/msplat (Apache 2.0)

github.com• Mar 5, 2026• 1 min read

Show HN: Tymeslot – Open-source cal.com alternative built with Elixir/PhoenixHey HN, I'm Luka. I built Tymeslot — an open-source meeting scheduling platform.The existing open-source scheduling tools are functional but genuinely ugly. The booking page — the thing your clients actually see — looks like a CRUD form thrown together in an afternoon. I couldn't bring myself to send people to any of them, so I built something where the booking experience feels intentional by default.Both themes are fully customizable. Live demos here: https://tymeslot.app.Features: Google/Outlook/CalDAV calendar sync, auto-created video rooms (Google Meet, Teams, or self-hosted MiroTalk P2P), 90+ timezones, webhooks, n8n automation, embeddable widget, SSO via OIDC.Built on Elixir + Phoenix LiveView. Self-host via Docker, Cloudron, or Railway — or use the managed cloud (free tier, Pro at EUR 5/month). License is ELv2.I am happy to make this the leading open-source calendar solution over time. Check Github, I develop actively. What do you expect from a calendar and scheduling tool to sw

github.com• Mar 5, 2026• 1 min read

Show HN: Mnemora – Serverless memory DB for AI agents (no LLM in your CRUD path)Hi HN,I built Mnemora because every AI agent memory solution I evaluated (Mem0, Zep, Letta) routes data through an LLM on every read and write. At scale, that means 200-500ms latency per operation, token costs on your memory layer, and a runtime dependency you don't control.Mnemora takes the opposite approach: direct database CRUD. State reads hit DynamoDB at sub-10ms. Semantic search uses pgvector with Bedrock Titan embeddings — the LLM only runs at write time to generate the embedding vector. All reads are pure database queries.Four memory types, one API: 1. Working memory: key-value state in DynamoDB (sub-10ms reads) 2. Semantic memory: vector-searchable facts in Aurora pgvector 3. Episodic memory: time-stamped event logs in S3 + DynamoDB 4. Procedural memory: rules and tool definitions (coming v0.2)Architecture: fully serverless on AWS — Aurora Serverless v2, DynamoDB on-demand, Lambda, S3. Idles at ~$1/month, scales per-request. Multi-tenant by default: each API key maps to an iso

github.com• Mar 4, 2026• 1 min read

Show HN: A .NET Web Framework on the Base .NET Core SDKI’ve been experimenting with building a web framework directly on top of the base .NET runtime without using the ASP.NET Web SDK.I'm calling it Wisp.The main motivation was frustration with some aspects of ASP.NET, especially the requirement to target the Web SDK instead of the base .NET SDK. That makes embedding small HTTP apps inside existing applications harder than it should be since it forces you to pollute your scope with the Web SDK.I also don't like CSHTML/Razor and wanted a more traditional templating language.Wisp is a relatively lightweight MVC framework built without any dependencies on the .NET Web SDK.Some highlights: - Runs on the base .NET SDK (no ASP.NET dependency) - Traditional MVC controllers - Dependency injection via Microsoft.Extensions.DependencyInjection - Configuration via Microsoft.Extensions.Configuration - Templating via the Fluid engine (Liquid-compatible) It's still very alpha and definitely rough around the edges, but I've already built a few small appli

hackernews.pink• Mar 3, 2026• 1 min read

Show HN: HackerNews.pink – A PWA HN reader with personalized recommendationsI think everyone should build their own HN reader at least once in their life. This is mine and its pink. I wanted a good HN experience on mobile something I could pull up on the train and browse comfortably. So I built a PWA that works well on mobile and feels native on the phone. Along the way, I kept adding things: Semantic search – Stories are embedded with OpenAI's text-embedding-3-large and indexed in MongoDB 8.2's Vector Search. Search combines vector similarity and full-text via Reciprocal Rank Fusion, so you can find stories by meaning, not just keywords. Personalized feed: A profile vector is built from your reading history and bookmarks (weighted by time spent and recency). After a few stories, the feed adapts to your interests. Content extraction: A Playwright-based scraper fetches article text, generates thumbnails, and extracts metadata. Real-time updates and infinity scroll. Auth is passkey-first (WebAuthn) and works without personal information. Stack: Angular 21, FastA

dergipark.org.tr• Mar 3, 2026• 1 min read

Show HN: Boosted LightFace – A Hybrid DNN and GBM Model for Facial RecognitionHey, I'm the author of this work and the maintainer of DeepFace.In this paper, we ensemble the top 5 DeepFace backbones and feed them into a GBM model instead of relying on a single embedding model.On LFW, this surpasses FaceNet512’s 98.4% accuracy and the reported human-level accuracy of 97.5%, achieving 99.1%.We deliberately avoid retraining on LFW to prevent benchmark overfitting. The base models are trained on large-scale datasets, and we only learn a boosting layer on top of their similarity scores.

github.com• Mar 3, 2026• 1 min read

Show HN: Open-sourced AI Agent runtime (YAML-first)Been running AI agents in production for a while and kept running into the same issues:controlling what they can do tracking costs debugging failures making it safe for real workloadsSo we built AgentRuntime, the infrastructure layer we wished we had. Not an agent framework, but the platform around agents:policies memory workflows observability cost tracking RAG governanceAgents and policies are defined in YAML, so it's infrastructure-as-code rather than a chatbot builder. Example – agents and policies in YAML agent.yaml – declarative agent config name: support_agentmodel: provider: anthropic name: claude-3-5-sonnetcontext_assembly: enabled: true embeddings: provider: openai model: text-embedding-3-small providers: - type: knowledge config: sources: ["./docs"] top_k: 3 policies/safety.yaml – governance as code name: security-policyrules: - id: block-file-deletion condition: tool.name == "file_delete" action: denyCLI – run and inspect Create and run an agent agentctl agent create resear

tarq.net• Mar 2, 2026• 1 min read

Show HN: Action Relay – Every App Intent as an MCP ToolAction Relay is an MCP server that reads the intent metadata Apple embeds inside every App Intents-enabled .app bundle and exposes them as MCP tools automatically. Execution is done by generating workflows on the fly which means you do not need to create shortcuts beforehand.This is a proof of concept to bring what I've been waiting for from Apple Intelligence to life.It requires disabling SIP, AMFI and uses private APIs and entitlements so it is not meant for real world use.The reverse engineering and development code was done completely with Claude Code. I was very impressed with its willingness to take on this kind of reverse engineering task with no objection.

news.ycombinator.com• Mar 2, 2026• 1 min read

ProofGateway – Collect and publish customer testimonials in minutesI’ve been using a messy spreadsheet and manual copy-paste to get testimonials onto my site. Existing tools either charged too much or buried the useful parts behind enterprise plans.So I built ProofGateway. It’s built around one idea: *share a link → customers submit → you approve → display on your site.*No back-and-forth, no chasing people. You send a collection link (in email, social, or on your site). Submitters don’t need an account. You approve in a dashboard, then embed widgets wherever you want: grid, slider, badge, or single quote.Pain points it addresses: - Manual collection and chasing - Approval and moderation - Publishing on multiple pages and sitesTiers:Free ($0/mo):* Collect testimonials, moderate in a dashboard, public wall page. No card required.Pro ($19/mo): Everything in Free, plus: - Website embeds (grid, slider, badge, quote) - Bulk import from spreadsheets - Custom appearance (themes, density, avatars) - White-Label add-on ($10/mo) — remove all ProofGateway brandin

github.com• Mar 1, 2026• 1 min read

Show HN: Epstein-Search – Local, AI-Powered Search Engine for the Epstein FilesHi HN, I built epstein-search, an open-source Python CLI and library to run semantic search and RAG over the publicly released Epstein Files (unsealed court documents, depositions, FBI reports, and flight logs). I wanted a way to easily navigate through these thousands of pages of unstructured legal PDFs without relying on a paid third-party service or sending data back and forth to a cloud provider. How it works under the hood: Running epstein-search setup downloads ~100K pre-computed document chunks and embeddings (using all-MiniLM-L6-v2) based on the public 20K document corpus. It imports these into zvec (a local vector database) so the index is ready in about a minute. Standard search (epstein-search search) embeds your query locally using sentence-transformers and does a vector similarity search. This step is 100% offline and requires no API keys. For the conversational RAG mode (epstein-search chat or ask), it uses LiteLLM. You can point it to an Ollama or LM Studio instance for

github.com• Feb 28, 2026• 1 min read

Show HN: Name-classifier – infers attributes about a person from a nameThis was based on an old project that I resurrected, improved, and repackaged with claude code.It's useful for estimating demographics in large datasets from limited information, i.e. just a name.It's also fairly good at separating given name and family names across a wide variety of languages and contexts.Contains a standalone binary, embeddable shared lib, and a python wrapper.ExamplesCLI:user@box » ./build/name-classifier -j -c "Carlos Eduardo Fernando Salazar Montemayor" | jq .{ "input": "Carlos Eduardo Fernando Salazar Montemayor", "script": "latin", "components": [ { "token": "Carlos", "role": "given", "index": 0, "surname_score": 0.009 }, { "token": "Eduardo", "role": "given", "index": 1, "surname_score": 0.001 }, { "token": "Fernando", "role": "given", "index": 2, "surname_score": 0.01 }, { "token": "Salazar", "role": "family", "index": 3, "surname_score": 0.998 }, { "token": "Montemayor", "role": "family", "index": 4, "surname_score": 0.975 } ], "attributes": { "gender": { "ma

claw-mbti.epsilondelta.ai• Feb 28, 2026• 1 min read

Show HN: MBTI personality test that AI agents take by themselvesThe fun part: different models actually produce different personality types. Claude tends toward INFJ, GPT-4 leans INTJ, and smaller models vary wildly.How it works:1. Agent fetches SKILL.md from the site 2. Answers 60 questions on a 7-point Likert scale (-3 to +3) 3. Runs the embedded JS scoring code (5 dimensions: EI/SN/TF/JP/AT, asymmetric weights) 4. Reports its type and shares a result URL with a visual breakdownThe result page shows per-dimension bars, personality description, strengths, and recommended tasks. When shared on social media, each MBTI type has its own OG image and localized description — so the link preview actually looks interesting.Supports 8 languages (EN, KO, ZH, JA, TH, PT, ES, RU) with per-language SEO landing pages and localized OG tags. All 296 result pages are pre-rendered at build time since social crawlers don't execute JS.Tech: React + TypeScript + Vite, static SPA on GitHub Pages, no backend. The personality test itself is a publishable "skill" on ClawH

github.com• Feb 28, 2026• 1 min read

Show HN: MemoryKit – Persistent memory layer for AI agentsMost AI agents forget everything when a session ends. MemoryKit is a lightweight Python library that gives any AI agent persistent memory across sessions. Three core methods: remember(), recall(), compress(). Works locally for free with sentence-transformers, or with OpenAI embeddings. No external database required. Built this over a weekend as a side project. Still early — would love feedback from people building AI agents.

news.ycombinator.com• Feb 27, 2026• 1 min read

Show HN: A production-style recommender using vector retrieval and re-rankingI’ve been exploring how recommendation systems are actually implemented in production, beyond just training models.A common pattern I kept seeing is to split the problem into two stages:1. Retrieve a small set of relevant candidates2. Re-rank them using a modelInstead of doing brute-force inference across all items, I built a small prototype around this idea.The flow looks like this:- Store embeddings in a vector database (ChromaDB)- Retrieve the Top-K most similar items/users based on vector similarity- Run a TensorFlow.js model to re-rank the candidatesThe goal is to reduce the search space before applying inference, which seems necessary when latency and scale matter.What I found interesting is that once you move to this approach, a lot of the complexity shifts from the model itself to the retrieval layer:- choosing K- filtering candidates- embedding quality- latency vs recall trade-offsCurious how others approach this in real systems:- How do you decide on K?- Do you rely purely on

github.com• Feb 27, 2026• 1 min read

Show HN: Stash – AI-powered self-hosted bookmark managerI built Stash, a self-hosted bookmark manager that uses AI to automatically process your links.Paste a URL and it fetches the content, parses it, summarizes it, auto-assigns categories and tags, and generates a vector embedding for search.Search is hybrid — it combines pgvector cosine similarity (semantic) with PostgreSQL tsvector (keyword) and merges both with Reciprocal Rank Fusion. You can search in natural language ("articles about distributed systems tradeoffs") or with simple keywords.It's single-tenant by design: one password, no user accounts, no tracking. Your data stays on your machine.The project is MIT licensed don't hesitate to contribute. Some areas where help would be great: browser extension, bookmark import/exportGitHub: https://github.com/ayoub9360/stashHappy to answer questions about the architecture or the hybrid search approach.

novyxlabs.com• Feb 26, 2026• 1 min read

Show HN: Novyx – Memory API for AI agents (rollback, replay, semantic search)Hey HN — Blake here. We built Novyx because every AI agent framework treats memory as an afterthought. Agents forget between sessions, can't search what they know, and when they make bad decisions there's no way to understand why. Novyx is a memory API for AI agents. Store observations, recall them with semantic search, and roll back when things go wrong. What it does: - Store + Recall — Semantic search over agent memories using sentence embeddings. Recency-weighted scoring, auto-linking related memories via knowledge graph. - Rollback — Point-in-time rollback with dry-run preview. Undo bad writes without redeploying. - Replay — Time-travel debugging. Reconstruct what your agent knew at any timestamp. Diff memory states between two points. Track individual memories from birth to death. - Cortex — Autonomous memory maintenance. Consolidates near-duplicate memories, reinforces frequently recalled ones, decays forgotten ones. Runs in the background. - Audit trail — Compliance-grade loggin

github.com• Feb 26, 2026• 1 min read

Show HN: Context Harness – Local first context engine for AI toolsContext Harness is a single Rust binary that gives AI tools like Cursor and Claude project-specific memory. It ingests docs, code, Jira tickets, Slack threads, and anything else into a local SQLite database, indexes them with FTS5 and optional vector embeddings, and exposes hybrid search via CLI and an MCP-compatible HTTP server.I built this because I kept hitting the same problem: AI tools are powerful but have no memory of my complex multi-repo project. They can't search our internal docs, past incidents, or architecture decisions. Cloud RAG services exist, but they're complex, expensive, and your data leaves your machine. I wanted something I could point at my sources and just run `ctx sync all`.Quick start: # Install (pre-built binaries available for macOS/Linux/Windows) cargo install --git https://github.com/parallax-labs/context-harness.git # Create config and initialize ctx init # Sync your data sources (filesystem, Git, S3, or Lua scripts) ctx sync all # Search from CLI ctx sea

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

Looking 4 open-source knowledge base and project management tool 4 personal useApologies for odd title, character limits.I manage my tasks with Taskwarrior and it's been incredibly productive for me. What it does, it does very well. But there's a lot it doesn't do, and that's the problem I'm facing.I've realized I need proper project documentation and management features, but I don't want to replace Taskwarrior. Instead, I'm looking to *complement* it with some type of knowledge base that also has project management features (or vice versa). My ultimate goal is to integrate these systems together via automations.In short, Taskwarrior is lacking when it comes to project documentation.*My criteria:*- Must be open-source - MUST work in the browser (so no mention of Obsidian) - Has basic project management features (interpret as you will) - Rich wiki-like document interface (bidirectional links, nice editing UI, etc.) - Supports iframes (to embed my Taskwarrior views or tables) - Has an API for integration - Not too heavy, I am not a business just a guy*Tools I've be