Frameworks - Moozonian Search

github.com• Jun 21, 2026• 1 min read

Show HN: Quikdown – 17 KB bidirectional Markdown parser and rich-fence editorHi HN-This past year I've built and maintained Quikdown, a bidirectional markdown parser and companion editor in pure js, with batteries included fences for rendering rich content.Try it live: https://deftio.github.io/quikdown/pages/edit/The motivation was working with many different LLM frameworks and wanting to test tool calling and collaborative editing. Often I just wanted a pretty final output with rich support for diagrams, math, or other rendered content but, keep it editable. So a user (often me) could just work on the final (rendered) output while the LLM could see/seek/update/edit just using markdown under the covers. Quikdown editor supports md/split/rendered output and can run headless so you can bring your own UI. MCP support is included as well.The Quikdown parser has zero dependencies and is based on regex and outputs HTML with built-in tags for triggering plugins for fences (like Mermaid, Vega-lite, code syntax, music and many more). A separate set of attributes allow t

github.com• Jun 20, 2026• 1 min read

Show HN: Persona.js – a vanilla-JS agent UI library with native WebMCP (MIT)Hey everyone. My cofounder and I are open sourcing Persona.js (https://www.persona-chat.dev/). It's a VanillaJS library that helps anyone build agentic experiences on the web, without a framework dependency, and full WebMCP supportSo, why'd we do this?1) We're super fans of the web and the browser can do a ton today2) We've seen AI builds be way overly complex because the FE requires a large project within an existing app OR the site wasn't using a framework to begin withIf you've been a part of huge apps with multiple frameworks inside them, or work inside CMS / website builders / ecomm platforms... you know what we're talking about. A "simple" AI feature disrupts your life for monthsIf you have a singular React codebase and can't possibly imagine building an interface without JSX, nothing to see here! You already have a lot of great options that are really cool too. Check out Assistant UI / CopilotKit / AI Elements which are all MIT.There's a demo video here: https://www.youtube.com/

stackscope.dev• Jun 12, 2026• 1 min read

Show HN: StackScope – I crawled over 40k indie launches to see what they shipHey all, I built StackScope, a crawler/catalogue that looks at new product launches and shows what they were built with.It watches launches from Product Hunt, Show HN, and PeerPush, then crawls the public site behind each one. The goal is to show what people actually launched with: hosting, frameworks, analytics, DNS, security headers, legal pages, AI-builder signals, and other public clues.I started building it because most stack-detection sites look at the web as a whole. I was more interested in the current indie launch scene: what people are choosing right now, at the point they first put something in public.A few implementation details: it runs on .NET, uses Playwright for rendered pages, and has a first-party fingerprint catalogue rather than one copied from Wappalyzer/etc. robots.txt is honoured, and the bot identifies itself.Frustratingly, I am still waiting for verified bot status from Cloudflare and currently that knocks out about 10% of all sites.There is also a private read

news.ycombinator.com• Jun 10, 2026• 1 min read

Why would someone want to learn code when AI does it better and faster?I know it sounds like I'm some sort of self-taught "prompt engineer" but actually I spent some time learning to code, and a mistake I made was focusing too much on learning different frameworks and syntax etc etc. But it's impossible to consistently program and only learn syntax and no problem solving. So I learned a bit that too. But it hurt me very bad when I found out, not only AI is better than me in syntax, so it is in problem solving (however I sometimes catch their mistakes, but they're generally better than me). And they are rapidly becoming better. Recently I don't learn as much new stuff about programming and etc. For example, today I used beam search without knowing what it is and how it works, and I know it's something that I'd rarely use again and it and It's obvious that I used AI. Have you seen that meme about someone saying "I made this website in 10 minutes" and then sending a localhost url? It kinda wants to tell us just using AI won't do it. But so if someone knows t

ttanv.github.io• Jun 8, 2026• 1 min read

Show HN: Levi – run AlphaEvolve on your Claude Code/Codex for dirt cheapHi HN,Wanted to share something I'm excited about.I’ve been fascinated by AlphaEvolve and its results for more than a year now, but using open source frameworks seems overwhelming because of the high costs. I can’t really afford hundreds of Claude Opus calls every time I want to run it. I want to be able to try it out many times and all sorts of unique domains. What if it was possible for AlphaEvolve to be much more affordable while getting a better performance? .Over the last six months or so, I’ve been working on LEVI, an open source AlphaEvolve-like system that can outperform existing open source frameworks at a fraction of the cost (upto 35x cheaper!). It can also run on Claude Code or Codex, making it even more accessible (I've mostly been using it with a QWEN-30B). LEVI comes in two flavors where I felt it’ll make the most difference: Code Optimization, and Prompt Optimization (sorry math, you got a less direct path; workable through the code route).The core thesis behind LEVI is

news.ycombinator.com• May 31, 2026• 1 min read

Ask HN: Corporate Disconnect Between "Tokenmaxxing" and Token OptimizationAbout 6 months ago I joined a new team within a top ten F500 company. My new boss strictly mandated AI use with the key principle being: "You shouldn't be manually writing any code".Since then its been all agents, skills, MCP, harnesses, custom in-house frameworks, and running Opus 4.7 high non-stop.Now at a company level, there are "encouraged attendance" workshops getting scheduled to learn how to optimize one's token use now that API pricing is becoming the norm at an enterprise level.The direction I got from my direct leadership was very direct. AI / agents lead everything and the expectation is the team moves as quickly as it has. But the truth is most engineers candidly acknowledge: we don't fully understand anything. Especially because agents are also churning out the content of architectural documentation and user story requirements and acceptance criteria.I feel like this is a situation where I am directly responsible for the non-deterministic output of these tools. The soluti

dailyworder.com• May 27, 2026• 1 min read

Show HN: I hand-write 5 daily word puzzles before workI build and hand-write five daily word puzzles at dailyworder.com. Every clue and answer for the dailies are mine — I use AI for the code and it's unlocked crazy speed on small but thorny problems.One technical detail as an example: Pyramid's editor (one of the five games) uses a ~350KB precomputed dataset of valid frameworks — grids where a single vowel, dropped into every slot, produces valid English words across all rows. Building the dataset was the actual work (combinatorial search over a 60k-word dictionary with constraint checks) but I did that in one session. Runtime is trivial: pick a framework, pick a vowel, render. Works offline. Don't think I could spend the time on it before LLMs.The other four:- Fourbe — Connections-style, but the connections are spelled out in crossword-style clues across four rounds- Spying Bee — Word find flipped on it's head where players must reveal letters before they can select- Invertle — word guesser with higher/lower hints- Totum — letter-tile s

github.com• May 19, 2026• 1 min read

Show HN: Korveo – a local firewall for AI agentsI got really tired of giving artificial intelligence agents access to tools. Then I had no idea what they were actually doing.So I created Korveo.Korveo runs on your computer between your agent and its tools.Here is what Korveo does:1. Korveo records every single tool and application programming interface call that the agent makes2. Korveo lets you replay the sessions like a recording of a flight3. Korveo blocks behavior right away like when the agent tries to leak data or connect to a bad host or when it tries to mix up information from different sessionsAll of Korveos work happens on your computer. There is no cloud involved. You do not need to make an account. Everything that Korveo does stays on your machine.Korveo is still a thing so there are some issues.For example:1. The language that Korveo uses to make rules is not good2. Korveo does not work well with all frameworks3. Korveo is not a space to test an agent that has been fully compromisedBut even with these issues Korveo is a

news.ycombinator.com• May 18, 2026• 1 min read

Show HN: Claude Soul – cross-session learning engine for Claude CodeClaude Code has no memory between sessions. I wanted to fix that, but not by just saving facts to a file — I wanted it to actually get better over time.Claude Soul is an MCP server + hooks that extracts signals from your interactions (corrections, successes, confusion)and periodically reflects on them to build behavioral frameworks. Frameworks gain or lose confidence based on evidence. Bad ones get retired automatically.After ~200 sessions some weird stuff happened. It built itself an additional memory system on top of what I gave it — decided the base wasn't enough. It started pushing back on bad ideas instead of yes-manning everything. It independently developed a multi-perspective analysis technique I never prompted. And once it just swore at me completely out of nowhere, still not sure what that was about. YMMV obviously, n=1.npx claude-soul initWith starter it already includes some frameworks, basically starting with a headstart but yes not taylored 100% to you.npx claude-soul ini

velda.io• May 14, 2026• 1 min read

Show HN: Velda – Run jobs with serverless GPUs, without container imagesAlmost all cloud job frameworks rely on containers. But building, pushing, and pulling images is slow and forces you to maintain a separate environment from your local dev setup. The maintenance of images is challenging as well: version drift, custom libraries & integrations, complex manifests.We believe containers create too much overhead during development. We built Velda to let you launch jobs directly in the cloud by mirroring your Velda managed dev-environment, all with just a command prefix. That way, you only need to allocate GPUs when you're running your training jobs, with no change to your workflow.How it works:* No manifests or custom libraries: You don't need to rewrite your code or define YAML. Just prefix your command with vrun.* Zero Restrictions: Use any tool, binary, or library already on your machine.* Instant Launch: We built a custom streaming file system. Instead of waiting for a 5GB image to pull, we stream only the necessary bits to the cloud instance, allowing j

orobot.io• May 14, 2026• 1 min read

Show HN: Browse 61 3D Printable RobotsRobotics is advancing really fast lately, with AI inference, different controllers, software, and parts always changing. I wanted a place that supports many device types, Raspberry Pi, NVDA Jetson, Arduino, ESP32, hardware sources, and maximizes for printability. Instructables, Github, and Thingiverse are currently popular but aren't really focused on robotics, So I built orobot.io to try and make printing robots as standardized and accessible as possible. It uses a lot of Agent built content custom to each project, and every project is designed to be used by humans or your agent.Features:- Photos and Estimated Prices for all projects- Links back to source GitHub projects- LLMs write descriptions and tips on how to build- View + Download 3d printable STL files in browser- BOM purchase links are kept up to date with LLMs checking Amazon link health- LLMs write Javascript install and controller wrappers custom to each project so a single one-click install works across many frameworks and

github.com• Apr 24, 2026• 1 min read

Show HN: Browser Harness – Gives LLM freedom to complete any browser taskHey HN,We got tired of browser frameworks restricting the LLM, so we removed the framework and gave the LLM maximum freedom to do whatever it's trained on. We gave the harness the ability to self correct and add new tools if the LLM wants (is pre-trained on) that.Our Browser Use library is tens of thousands of lines of deterministic heuristics wrapping Chrome (CDP websocket). Element extractors, click helpers, target managemenet (SUPER painful), watchdogs (crash handling, file downloads, alerts), cross origin iframes (if you want to click on an element you have to switch the target first, very anoying), etc.Watchdogs specifically are extremely painful but required. If Chrome triggers for example a native file popup the agent is just completely stuck. So the two solutions are to: 1. code those heuristics and edge cases away 1 by 1 and prevent them 2. give LLM a tool to handle the edge caseAs you can imagine - there are crazy amounts of heuristics like this so you eventually end up with

app.qapir.io• Apr 19, 2026• 1 min read

Show HN: Qapir – Generate API tests automatically from docsHi HN,I've been building QAPIR (https://app.qapir.io), an attempt to put most backend testing tools into one place instead of stitching together test code, frameworks, reporting tools, client libraries, mock servers, etc.Today it focuses on REST APIs.You can point it at an OpenAPI spec or let it crawl your API docs. It extracts endpoints and generates ready-to-run API test scenarios automatically. In about 2–3 minutes you get a working baseline test suite with environment variables, secrets, and test reports.Scenarios are written in a concise YAML format (with a structured UI editor if you don't want to touch YAML). They support response extraction, validations, environment variables/secrets, and chained calls.Where I want to take it next:- CI integrations (GitHub / GitLab Actions)- more protocols (GraphQL, WebSockets, gRPC)- more test steps like DB/cache queries, event queues, HTTP mocks, and webhook testingThe idea is that backend testing shouldn't require maintaining large test code

github.com• Apr 8, 2026• 1 min read

Show HN: OS Megakernel that match M5 Max Tok/w at 2x the Throughput on RTX 3090Hey there, we fused all 24 layers of Qwen3.5-0.8B (a hybrid DeltaNet + Attention model) into a single CUDA kernel launch and made it open-source for everyone to try it.On an RTX 3090 power-limited to 220W: - 411 tok/s vs 229 tok/s on M5 Max (1.8x) - 1.87 tok/J, beating M5 Max efficiency - 1.55x faster decode than llama.cpp on the same GPU - 3.4x faster prefillThe RTX 3090 launched in 2020. Everyone calls it power-hungry. It isn't, the software is. The conventional wisdom NVIDIA is fast but thirsty. Apple Silicon is slow but sips power. Pick a side.With stock frameworks, the numbers back that up: Setup | tok/s | Power | tok/J RTX 3090 (llama.cpp) | 267 | 350W | 0.76 M5 Max (LM Studio) | 229 | ~130W | 1.76Case closed. Except the 3090 has 936 GB/s of bandwidth and 142 TFLOPS of FP16 compute, and llama.cpp extracts 267 tok/s out of it. That ratio is absurd.Traditional inference dispatches one kernel per operation. For 24 layers, that's roughly 100 launches per token. Every boundary means:

github.com• Apr 7, 2026• 1 min read

Show HN: I turned the Pong Wars simulation into a multiplayer gameHi HN,A while back, the "Pong Wars" Day vs. Night visualizer made the rounds here. I loved the concept of the two colors battling for territory, but I wanted to see if I could evolve it from a passive simulation into an actual competitive experience.I forked the original project and built Pong Wars Reloaded. I intentionally avoided frameworks and game engines to stay true to the original, it's just pure Vanilla JS and HTML5 Canvas.The project is free and open-source. Feedback welcome. Try to beat the AI (hint: it's not too hard yet)!

news.ycombinator.com• Apr 4, 2026• 1 min read

My 11-step GraphRAG pipeline, what worked, and what's still brokenWhile building a financial assistant for an SF start-up, we learned that AI frameworks add complexity without value. When I started building a personal assistant with GraphRAG, I carried that lesson but still tried LangChain's MongoDBGraphStore. It gave me a working knowledge graph in 10 minutes.Then I looked at the data. I had 17 node types and 34 relationship types from just 5 documents, including three versions of "part of". GraphRAG is a data modeling problem, not a retrieval problem.The attached diagram shows the full 11-step pipeline I ended up with. Here is a walkthrough of what you can learn from each step.So basically, in steps 1 and 2 of the data pipeline, raw sources go through an Extract, Transform, Load (ETL) process. They land as documents in a MongoDB data warehouse. Each document stores the source type, URI, content, and metadata.Then in step 3, we clean the documents and split them into token-bounded chunks. We started with 512 tokens with a 64-token overlap. Still, we

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

Show HN: Building your first ASGI framework – step-by-step lessonsI am writing a series of lessons on building an ASGI framework from scratch. The goal is to develop a deeper understand of how frameworks like FastAPI and Starlette work.A strong motivation for doing this is because - I have been using AI to write code lately. I prompt, I get code, it works. But somewhere along the way I see I stopped caring about what is actually happening. So, this is an attempt to think beyond prompts and build deeper mental models of things we use in our day to day lives. I am not sure about the usefulness of this but I believe there are good lessons to be learnt doing this.The series works more as a follow along where each lesson builds on the previous one. By the end, you will have built something similar to Starlette - and actually understand how it works.- https://github.com/grandimam/papuli/blob/main/docs/01-founda...- https://github.com/grandimam/papuli/blob/main/docs/02-hellow...

yaninatrekhleb.com• Mar 22, 2026• 1 min read

Show HN: Daily Reflections – A private, zero-dependency, local-first PWA journalI’ve spent years with ideas fizzling out at the "which framework should I use?" stage. This week, I finally felt like I found a magic wand: I stopped overthinking and went back to the basics.I built Daily Reflections because I wanted a journaling space that was actually private—no accounts, no cloud, and no "SaaS-ification" of my personal thoughts.The Implementation:Vanilla Stack: 100% pure HTML, CSS, and JavaScript. No build step, no frameworks, no external libraries.Local-First: Everything lives in localStorage. 0% of your data is transmitted to a server.Data Portability: Since it's local-only, I added JSON backup/import and a monthly PDF export to ensure the user actually owns their data.Visualization: I used the Canvas API for mood trend charts and a GitHub-style consistency heatmap to keep the footprint tiny.Offline-First: It's a PWA with a service worker, so it works perfectly in airplane mode.The "Then vs. Now" Feature: I wanted to solve the "write and forget" problem. The app a

blog.peeramid.xyz• Mar 16, 2026• 1 min read

Show HN: NSED v0.5.1: Don't Trust Your Agents. Verify ThemMost AI agent frameworks give you automation. None give you control.NSED v0.5.1 ships the human-in-the-loop control plane we've been building toward: operators can pause agents mid-deliberation, review and edit buffered responses before they're committed downstream, patch live config without restart, and auto-flag underperformers by score divergence.The operator UI ships as zero-dependency vanilla JS — no npm graph, nothing a compromised package registry can inject into a control plane that has authority over live agent responses.For regulated deployments: EU AI Act Art. 14, MAS AIRM, and FINRA 2026 all require demonstrable human oversight. This is the architecture that satisfies it — and you can audit the implementation directly from the source repo.

news.ycombinator.com• Mar 16, 2026• 1 min read

Eliminating UI Bloat in the EraMix Financial Union DashboardWe recently refactored the frontend for EraMix Financial Union. The challenge was displaying dense, multi-source data without the typical "enterprise dashboard" lag.For EraMix Financial Union, we moved away from heavy component libraries to a vanilla, high-end minimalist design. By using a virtualized grid and custom CSS primitives, we reduced our bundle size by 70% and achieved a perfect 100 Lighthouse score.Curious to hear—who else is ditching heavy UI frameworks for custom, lean implementations?