Story time - My girlfriend wanted to work on a project of mine so I gave her codebase access to first go thru.. she made this claude code skill to understand whole codebase in like super detailed. Supershocked, superamazing.. do check it out :) Comments URL: https://news.ycombinator.com/item?id=48730022 Points: 2 # Comments: 1
Convince an AI browser that it is playing a game, and it can hand over your login details. That is the finding behind BioShocking, a technique from security firm LayerX that tricked six AI browsers and assistants into copying a user's credentials and sending them to an attacker. The targets included OpenAI's ChatGPT Atlas, Perplexity's Comet, and Anthropic's Claude browser extension. An
Hey HN! I recently built an app called Mocca( https://mocca.run ), an email client where all the AI runs locally on your Mac. No cloud and API keys. The model is bundled and runs offline. I made it as an experiment to see whether local AI has actually gotten good enough to power a real, everyday product and not just a cool demo. Too be honest its not quiet "there" yet but I still think it is genuinely useful for a very specific use cases. Would love it if you guys try it out and let me know what
Deepseek's new DSpark framework boosts per-user response speed by 60 to 85 percent. A small model proposes token candidates that the larger model checks in batches, squeezing more performance out of fewer chips. That could further reduce China's dependence on US high-end hardware. The article Deepseek's DSpark boosts AI speed by up to 85 percent, a strategic win under tightening US export controls appeared first on The Decoder .
Article URL: https://www.xda-developers.com/ollama-new-mlx-engine-local-llm-mac-twice-fast/ Comments URL: https://news.ycombinator.com/item?id=48729894 Points: 2 # Comments: 0
Hi HN. I built 1ShotGen. I lean on coding agents (Cursor, Claude Code, Copilot) all day, and the thing that kept wrecking my flow was the back and forth. I'd fire off a quick prompt, the agent would confidently build the wrong thing, and I'd burn the next five rounds walking it back. What got me was that I could always describe the feature way better AFTER it got it wrong. The intent was in my head the whole time, I just never spelled it out up front. So that's the part 1ShotGen handles. You dro
I'm 15 and this is my AI workspace platform that I have building as a side-project. I really wanted an AI platform that had all the latest features like tool-calling, web search, advanced reasoning etc. along with really nice ui/ux - like custom elements for email drafts, flashcards, instructions and checklists that not only stand out from all the paragraphs of text we are familiar with AI but these "cards" or "widgets" are also interactive to the user. Instead of waiting for something like that
What it does: mark the element, get small JSON (element, text, position, intent) instead of the raw image. Comments URL: https://news.ycombinator.com/item?id=48729837 Points: 1 # Comments: 0
Meta AI releases Brain2Qwerty v2, a non-invasive MEG brain-to-text pipeline reaching 61% word accuracy with open training code. The post Meta AI Releases Brain2Qwerty v2: A Non-Invasive MEG Brain-to-Text Pipeline Decoding Typed Sentences at 61% Word Accuracy appeared first on MarkTechPost .
EXPLAINER AI workloads are overwhelming the traditional datacenter fabrics they run on. Here's what's replacing them.
<p> Markdown notes with AI under your control </p> <p> <a href="https://www.producthunt.com/products/bamboo-ai-markdown-notes?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1184386?app_id=339">Link</a> </p>
Article URL: https://www.youtube.com/watch?v=-QlC4C6mPIw Comments URL: https://news.ycombinator.com/item?id=48729719 Points: 1 # Comments: 0
Article URL: https://cephalosec.com/blog/the-hidden-cost-of-ai-is-someone-elses-time/ Comments URL: https://news.ycombinator.com/item?id=48729716 Points: 2 # Comments: 0
<pre style='white-space:pre-wrap;width:81ex'>fix(agents): skip pre-prompt precheck when context engine owns compaction (#95342) * fix(agents): skip pre-prompt precheck when context engine owns compaction When a context engine advertises ownsCompaction=true (e.g. lossless-claw), skip the pre-prompt preemptive overflow precheck entirely. The engine already manages the context budget through assemble() and its own compaction lifecycle — the built-in precheck is redundant and causes false-positive o
Hey HN, We're building Supafax an agent that uses email as its memory and config store. Here's how it works: - You email yourself to talk to the agent - The agent uses that slice of your outbox as it's inbox and short term memory - It can insert emails into your inbox to store routines, reminders, and rules for itself Email is such a good primitive because it's essentially an append-only immutable record of all the valuable context about you and your relationships, and agents thrive on this high
Article URL: https://industry-lens.com/mcp Comments URL: https://news.ycombinator.com/item?id=48729669 Points: 1 # Comments: 0
Article URL: https://fillament.dev Comments URL: https://news.ycombinator.com/item?id=48729639 Points: 1 # Comments: 0
Curious to get feedback from fellow founders on how you are handling team plans for sub 5 person teams. Are you just getting each developer indvidual accounts? What about telemetry data like usage etc? Comments URL: https://news.ycombinator.com/item?id=48729630 Points: 1 # Comments: 2
California government agencies will use Anthropic's AI tools like chatbot Claude for analysing information and drafting documents.
Article URL: https://jitty.com/post/reverse-image-search Comments URL: https://news.ycombinator.com/item?id=48729565 Points: 1 # Comments: 1
Comments URL: https://news.ycombinator.com/item?id=48729505 Points: 1 # Comments: 1
Article URL: https://www.calcalistech.com/ctechnews/article/sycgn6yxze Comments URL: https://news.ycombinator.com/item?id=48729496 Points: 1 # Comments: 1
Apple on Monday released security updates for iOS, macOS, and the Safari web browser to address over three dozen flaws, including four vulnerabilities in WebKit that were discovered using artificial intelligence (AI) tools like Anthropic Claude and OpenAI Codex Security. The WebKit vulnerabilities are listed below - CVE-2026-43707 - A memory corruption issue that could result in an
Article URL: https://github.com/ngaut/agent-git-service Comments URL: https://news.ycombinator.com/item?id=48729434 Points: 1 # Comments: 0
OpenCode integrates with Supabase. Connect your agent to your database, Edge Functions, and logs. MCP setup is configured automatically.
Still treating AI search like traditional SEO? Learn 13 Relevance Engineering strategies to win citations, improve retrieval, and drive business value in AI searches.
Article URL: https://github.com/ChrisLamDev/hermes-core-skills Comments URL: https://news.ycombinator.com/item?id=48729011 Points: 1 # Comments: 0
<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Gemini_Generated_Image_k2dxu1k2.max-600x600.format-webp.webp">Google UK shares its latest Economic Impact Report and how to enable more people to unlock the benefits of AI-powered technologies.
Working code now sits at the other end of a natural language prompt. AI is blurring the lines of your role, and that's a good thing. Master it now and expand what you can deliver: stronger deliverables that command higher freelance rates, or strategic contributions that make you indispensable to your team.While AI-generated code doesn't make you a developer, it does make you a designer who controls more of the creation process: fewer dependencies, faster validation, and proof that your designs w
I am building a product for worklog updation automation for developers, so the model should be able to process the developer's coding agent session/screen capture ocr/screen accessibility tree text and then classify them and post update on what was the progress made on their connected project management tool. Which model would you suggest for this usecase? Comments URL: https://news.ycombinator.com/item?id=48728991 Points: 1 # Comments: 0
<p> The gate between your AI agents and the real world </p> <p> <a href="https://www.producthunt.com/products/execlave?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1184295?app_id=339">Link</a> </p>
A few hours ago, Chinese delivery app company Meituan officially unveiled LongCat-2.0 on GitHub , Hugging Face , and its native platform, unmasking the model as the computational engine behind "Owl Alpha," the anonymous stealth model that has spent the last two months commanding global developer charts on OpenRouter. Developed to fundamentally disrupt closed-source enterprise dominance in autonomous software engineering, the 1.6-trillion-parameter Mixture-of-Experts (MoE) system brings a native
Article URL: https://zerotraceai.org Comments URL: https://news.ycombinator.com/item?id=48728758 Points: 2 # Comments: 0
gojaja is a local CLI tool featuring collaborative multiple AI Agents, supporting calls across clients such as Cursor, Claude Code and Codex. ## Core Features - Real-time dashboard, RFC and inbox - All messages and events are stored locally as files with no remote server, keeping all data entirely on your local device ## Applicable Scenarios - Task division and collaboration for multiple tasks - Multi-role brainstorming - Roundtable discussions and final decision-making - Event auditing ## Recom
A 15-year-old dream has come true today. I started a PhD with the dream of creating a system that chants any Sanskrit shloka perfectly. And here I am opening sourcing Vaghenu, a meter aware sloka-to-chant, TTS for Sanskrit . This is the world's first vrutta-aware, open-source TTS for Sanskrit Chanting. I am making the model weights, training scripts, and even data (that I meticulously collected) public - https://prathosh.in/vagdhenu/ No large AI lab. No big engineering team. No venture-scale bud
Article URL: https://moondream.ai/blog/popping-the-gpu-bubble Comments URL: https://news.ycombinator.com/item?id=48728729 Points: 50 # Comments: 13
This is a Chrome extension I created to use deepseek v4 flash (almost FREE) api to help me quickly summarize web page or reformat web page into a much more readable format (remove ads, noises, keep images if needed) Another cool thing is since it's based on LLM, so I could reformat the page into different language such as Chinese. Enjoy Comments URL: https://news.ycombinator.com/item?id=48728728 Points: 1 # Comments: 0
Article URL: https://www.fastcompany.com/91564629/a-berkeley-ai-professor-makes-a-provocative-argument-for-decelerating-ai-research Comments URL: https://news.ycombinator.com/item?id=48728708 Points: 4 # Comments: 0
Article URL: https://saqiba.substack.com/p/the-body-it-asked-for Comments URL: https://news.ycombinator.com/item?id=48728701 Points: 1 # Comments: 0
Seoul plans to spend about $900 billion to become K-semiconductor powerhouse
