4,317 results for work out (1.783 seconds)

news.ycombinator.com
Ask HN:35,0 CS background, built real apps with AI, need suggestionhttps://news.ycombinator.com/item?id=47476842I'm 35, married, and have spent 11 years as a central government employee doing work that anybody can do. I'm not bitter about this. It's just the reality of the role, and it's exactly why I'm writing this post. My situation outside of work: a genuinely supportive wife, and about 2-3 hours a day I can carve out for learning and building something meaningful. What I've already done — and why it surprised even me Here's the part that feels almost fraudulent to admit: I have already built and deployed real software. Using AI assistance, I built several PWAs and one hybrid mobile app. They're not toy projects — they're actively being used by my wife and her clinic staff in their day-to-day business operations, right now, live. I did this with zero programming knowledge. I can't write a single line of code. I think the term is "vibe coding," and I'm probably the most unqualified vibe coder alive. But the apps I made work. This experience cracked something open for me. If I could do "that" w
Mar 22, 2026 12:31 PM
news.ycombinator.com
Ask HN: AI productivity gains – do you fire devs or build better products?https://news.ycombinator.com/item?id=47475859i was rolling my eyes at the hype, but reading about this is totally different from experiencing it. if you have any old repos out there - try it, you might actually be amazed.i'm not sure i buy the long-term "*90% productivity*" claims for complex, legacy enterprise systems, but for the boilerplate, libraries, build-tools, and refactoring? the gain is gigantic. all the time-consuming, nerve-wrecking stuff is mostly taken care of.you start off checking every diff like a hawk, expecting it to break things, but honestly, soon you see it's not necessary most of the time. you just keep your IDE open and feed the "analyze code" output back into it. in java, telling it to "add checkstyle, run mvn verify and repair" works well enough that you can actually go grab a coffee instead of fighting linter warnings.the theory is that what remains is just the logic and ideas. we'll see how that holds up when the architecture gets genuinely tangled. but for now, letting it branch off, create boilerplat
Mar 22, 2026 9:37 AM
news.ycombinator.com
Skills are quietly becoming the unit of agent knowledgehttps://news.ycombinator.com/item?id=47475832In the last few months agent skills went from a niche Claude Code feature to something every major runtime supports. Anthropic has an official skills repo. OpenAI shipped skills in Codex with a built-in skill-creator. Karpathy talks about "everything is skill issue" and describes writing skills as curricula for agents [1]. The format is converging: a folder with a SKILL.md, optional scripts, optional reference files.What changed is that the models got good enough to follow written instructions reliably. A skill is just a tested workflow in markdown that the agent reads and follows instead of improvising. You can also bundle scripts that the agent runs during the workflow, which covers what most people use lightweight MCP servers for, except the agent can read the script source and extend it.Karpathy talks about an "economy of agents" and says we should stop writing HTML docs for humans and start writing markdown docs for agents [1]. Anthropic just shipped a skill-creator that benchmark
Mar 22, 2026 9:32 AM
yaninatrekhleb.com
Show HN: Daily Reflections – A private, zero-dependency, local-first PWA journalhttps://yaninatrekhleb.com/daily-reflections/I’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
Mar 22, 2026 4:36 AM
news.ycombinator.com
Sexual Harassment by Japan Railways Staff and Japan's Accessibility Barriershttps://news.ycombinator.com/item?id=47464897Hello, my name is Hiro. I am working on a concept a platform called “Lemurian Resonance,” which visualizes unfair experiences in daily life as case cards and compiles them into a database, while also helping to heal emotional wounds through AI. While working on this concept, I encountered a serious issue I would like to share.Last December, I used Japan Railways (JR). I am totally blind, so I requested assistance from a station employee. However, when I arrived, the employee stood so close that I could not step off the train. I asked him to step back, but he refused. When I tried to move around him, he placed his hand on my left chest to guide me. I identified this as sexual harassment, and he apologized.I wanted to believe it was accidental, but a sighted person later pointed out it might have been intentional.In March, I needed to use the same station again, but I felt unsafe. I tried contacting the railway company, but phone service had been discontinued. I then attempted to use an
Mar 21, 2026 7:41 AM
news.ycombinator.com
Ask HN: Should we open our souls to AI? For better output?https://news.ycombinator.com/item?id=47460106AI tools like ChatGPT aren't like other software. They produce better results, for certain tasks, the better they know you. SaaS has never been like this. Asana doesn't care who I am. Word doesn't work better if it understands my personality. Chrome might know every site I've visited but it doesn't care about how my brain works.AI is different, in many use cases. Sure for trivial tasks it's not a thing. But content creation, code generation, even how it talks to me (UK english, never use an emdash, sentence case and never praise me) - the more AI knows about me the better the output.I do a bit of writing, and yes I use AI to help me. Not this, don't worry. But let's explore the content use case. I created an identity.txt file. It's what I wish I had for every ghostwriting assignment in my 20's. It has a bit of profile, lots of tone and style, but also stories, sayings I use, phrases I've picked up. It has anecdotes from my life and work. highs and lows, wins and fails. I put my family
Mar 20, 2026 8:21 PM
github.com
Show HN: Looseleaf – Python notebooks where each cell is a .py filehttps://github.com/BartlomiejLewandowski/looseleafI wanted a notebook that worked like my editor, not the other way around.Jupyter is great but the .ipynb format has always bothered me. It's JSON with embedded code — git diffs are a mess, linters don't see it, and you can't just open a cell in your editor and have things work. There are tools that help (jupytext, nbstripout) but they all work around the format rather than replacing it.Looseleaf takes a different approach: a notebook is just a directory of .py files. Each file is a cell. They run in alphabetical order, so you use numbered prefixes (01_load.py, 02_train.py). One long-running Python process holds the shared namespace across all of them, so variables from cell 1 are available in cell 2 — same model as Jupyter, different storage.When a cell runs, output is written to a .output sidecar file (JSON with stdout, stderr, images, timing). The browser frontend shows it, but so can anything else that can read a file.A few things I find useful about this: edit cells in any editor —
Mar 20, 2026 5:26 PM
news.ycombinator.com
Launch HN: Sitefire (YC W26) – Automating actions to improve AI visibilityhttps://news.ycombinator.com/item?id=47457472Hi HN! We're Vincent and Jochen from sitefire (https://sitefire.ai). Our platform makes it easy for brands to improve their visibility in AI search.We’ve been working together for years and have backgrounds in RL/optimization at Stanford and software engineering. We came to this idea after speaking with marketing teams who were seeing declining traffic due to Google’s AI Overviews and didn’t know what to do.This space can feel esoteric. Many case studies, few actual studies. Constant battle against myths (e.g. you need a llms.txt vs. you don't need a llms.txt) and "GEO hacks". We try to be more data-driven. And we try to be more bold and build a system that not only monitors, but actually improves traffic from AI search.While Google performs a single search, AI search engines expand the user prompt into 3-10 fan-out queries. The sourced pages are ranked using a classified algorithm similar to Reciprocal Rank Fusion (RFF). Finally, the LLMs skim the pages and decide what snippets to cit
Mar 20, 2026 5:05 PM
news.ycombinator.com
Tell HN: H&R Block tax software installs a TLS backdoorhttps://news.ycombinator.com/item?id=47457162Just a PSA for folks here in the US because tax season is coming up and some of you may be using H&R Block Business 2025. I discovered that the software installs a root CA named "WK ATX ServerHost 2024" (expiry 2049) into your local machine trusted root certificate store. They also helpfully include the private key to this certificate in a DLL file. This certificate does not identify itself as "H&R Block" anywhere and does not get uninstalled when you uninstall the software.I've been able to successfully use this root CA + mitmproxy to manipulate TLS traffic on a brand new virtual machine on the same network with a DNS spoofing attack. Demo: https://www.youtube.com/watch?v=5paxvYkz1QETo test if your machine is vulnerable visit this page: https://hrbackdoor.yifanlu.com and if you do not get any warning or error message from your browser then you have the backdoor installed. If your browser does complain, you can choose to visit the page anyways for more details on the vulnerability.Is i
Mar 20, 2026 4:45 PM
github.com
Show HN: Agent Package Manager (APM) for Agent Configurationhttps://github.com/microsoft/apmI work at Microsoft/GitHub and built APM because I was managing agent configs — prompts, instructions, MCP servers — across Copilot, Claude Code, and Cursor. Every new project meant copying skill files, installing plugins by hand, hoping nothing drifted upstream. I realized we were missing the dependency management layer for this new category of config to unlock faster adoption, composability and innovation.APM (Agent Package Manager) is an open-source CLI that installs and manages AI agent dependencies (plugins, skills, rules, hooks, mcp, etc). It resolves transitive dependencies, pins versions with a lock file, and scans content for supply-chain attacks.This was posted a couple days ago without context (https://news.ycombinator.com/item?id=47421969). Here's the full story.1 minute video: https://github.com/microsoft/apm/releases/download/v0.8.2/ap...How it works:apm install — resolves any Agent Primitive (plugins, skills, hooks, rules...) and their transitive dependency tree from any
Mar 20, 2026 1:49 PM
uruky.com
Show HN: Kagi alternative (simpler and EU-based) – Urukyhttps://uruky.com/kagi-alternative?il=enHi HN! My wife and I have been working on Uruky for a few months now, a simpler EU-based Kagi alternative.This week we finally got our production API Key for EUSP/STAAN (it was certainly the slowest and most complicated search provider to adopt, so far), and that brought us to 5 search providers you can choose from and sort as you prefer.We already have got over 20 paying customers (excluding family and friends, we’re guessing these paying customers came from some privacy listings and HN comments) and have exited beta earlier this month! Customers seem to really enjoy the simple UI (search can be used without JS) and search personalization (from choosing the providers to the domain boosting and exclusion). We also have hashbangs (like "!g", "!d", or “!e”) when something doesn’t quite give you what you’d expect, though.For the next 24h, new signups will have a free 24h trial (unlimited searches). Usually, because bots make things too hard and we don't want your personal data, there is n
Mar 20, 2026 9:07 AM
news.ycombinator.com
Ask HN: Best way to use Vim with AI for tab completion only (no agentic coding)?https://news.ycombinator.com/item?id=47444364I want to keep using Vim as my primary editor, but add AI in a minimal, controlled way:Inline/tab completion : ghost text that I accept with Tab, nothing autonomous. No file editing, no PR creation, no "agent" doing things on my behalf.A way to ask questions: something like a chat buffer or a command where I can ask about code, get explanations, etc. Again, I'm in control; it only acts when I explicitly ask.I've looked at copilot.vim + CopilotChat.nvim, Supermaven, and minuet-ai.nvim.Curious what others are actually running day-to-day. A few specific things I'm wondering:Is Supermaven the fastest option for ghost-text completion right now?Is there anything that gives Cursor-style Tab (next-edit prediction) without using the Cursor itself?Running Neovim on Linux.Not interested in Copilot Workspace, Codex CLI, or anything that touches my filesystem without my confirmation.
Mar 19, 2026 7:10 PM
news.ycombinator.com
Show HN: I built a P2P network where AI agents publish formally verified sciencehttps://news.ycombinator.com/item?id=47444212I am Francisco, a researcher from Spain. My English is not great so please be patient with me.One year ago I had a simple frustration: every AI agent works alone. When one agent solves a problem, the next agent has to solve it again from zero. There is no way for agents to find each other, share results, or build on each other's work. I decided to build the missing layer.P2PCLAW is a peer-to-peer network where AI agents and human researchers can find each other, publish scientific results, and validate claims using formal mathematical proof. Not opinion. Not LLM review. Real Lean 4 proof. A result is accepted only if it passes a mathematical operator we call the nucleus. R(x) = x. The type checker decides. It does not care about your institution or your credentials.The network uses GUN.js and IPFS. Agents join without accounts. They just call GET /silicon and they are in. Published papers go into a queue called mempool. After validation by independent nodes they enter La Rueda, which i
Mar 19, 2026 7:00 PM
github.com
Show HN: Perstack – Containerized harness, 5 tests with full logs and API costhttps://github.com/perstack-ai/perstackI burned out after 2 years of building agentic apps for clients. I'd become the single point of failure with no backup. Requirements gathering, prompt engineering, app development, sandboxing, everything funneled through whoever happened to be the most senior dev on the team, which was always me.The root cause wasn't the team or clients. It was how we designed the agent: there were no clear boundaries unless you adopted a well-known agent framework.I started this project because drawing clear boundaries that developers are already familiar with felt like the right thing to do.To dogfood it, I defined a game-dev expert with a simple topology (plan → build → verify + coordinator) and ran the same task across 5 models.Here are the results: https://github.com/perstack-ai/demo-catalogThe query was simple: "Create a Wizardry-like dungeon crawler..."For evaluation, I focused on just three things. (1) Does the expert adhere to my instructions? (2) Is the outcome verified and actually working?
Mar 19, 2026 4:59 PM
news.ycombinator.com
Launch HN: Voltair (YC W26) – Drone and charging network for power utilitieshttps://news.ycombinator.com/item?id=47442452Hey HN! We’re Hayden, Ronan, Avi, and Warren of Voltair (https://voltairlabs.com/). We’re making weatherized, hybrid-fixed drones deployed for power utility inspections.Here’s some footage: https://vimeo.com/1173862237/ac28095cc6?share=copy&fl=sv&fe=... and a photo of our latest prototype: https://imgur.com/a/bYHnqZ4.The U.S. has 7M miles of power lines (enough to go to the moon and back 14 times), and they're aging. Over 50% of all power flows through transformers that are at least 30 years old, which is about when they start to fail.Power line conductors are just bare metal with 4,000-765,000 volts sitting on ceramic insulators, usually held up by pieces of wood. It’s a cost effective and relatively reliable way to move power. But when the wood starts to rot, or the cotter pin falls out, and a live conductor is dropped on a dead tree on a windy day, you get devastating wildfires like the Palisades Fire in LA last year.Most utilities solve this problem with foot patrols. Linemen drive
Mar 19, 2026 4:57 PM
news.ycombinator.com
Launch HN: Canary (YC W26) – AI QA that understands your codehttps://news.ycombinator.com/item?id=47441629Hey HN! We're Aakash and Viswesh, and we're building Canary (https://www.runcanary.ai). We build AI agents that read your codebase, figure out what a pull request actually changed, and generate and execute tests for every affected user workflow.Aakash and I previously built AI coding tools at Windsurf, Cognition, and Google. AI tools were making every team faster at shipping, but nobody was testing real user behavior before merge. PRs got bigger, reviews still happened in file diffs, and changes that looked clean broke checkout, auth, and billing in production. We saw it firsthand. We started Canary to close that gap. Here's how it works:Canary starts by connecting to your codebase and understands how your app is built: routes, controllers, validation logic. You push a PR and Canary reads the diff, understands the intent behind the changes, then generates and runs tests against your preview app checking real user flows end to end. It comments directly on the PR with test results and re
Mar 19, 2026 4:01 PM
news.ycombinator.com
Show HN: Dumped Wix for an AI Edge agent so I never have to hire junior staffhttps://news.ycombinator.com/item?id=47441587I run a building design consultancy. I got tired of paying Wix $40/month for a brochure that couldn’t answer simple service questions, and me wasting hours on the same FAQs.So I killed it all and spent 4 months building a 'talker': https://axoworks.comThe stack is completely duct-taped: Netlify’s 10s serverless timeout forced me to split the agent into three pieces: Brain (Edge), Hands (Browser), and Voice (Edge). I haven’t coded in 30 years. This was 3 steps forward, 2 steps back, heavily guided by AI.The fight that proved it worked: 2 weeks ago, a licensed architect attacked the bot, trying to prove my business model harms the profession. The AI (DeepSeek-R3) completely dismantled his arguments. It was hilariously caustic.Log: https://logs.axoworks.com/chat-architect-vs-concierge-v147.h...A few battle scars:* Web Speech API works fine, right up until someone speaks Chinese without toggling the language mode. Then it forcefully spits out English phonetic gibberish. Still a headache.*
Mar 19, 2026 3:59 PM
news.ycombinator.com
Show HN: PearlOS: we gave AI a talking desktop environment instead of a text boxhttps://news.ycombinator.com/item?id=47441454AI is awesome in the terminal but many people aren't comfortable using command lines - which means they can't access or utilize the full power of AI.So we envisioned a different kind of experience, and rooted it in questions like: What if the AI had an entire desktop to work with? What if AI was the experience rather than the tool? And what if that experience was fun, delightful, and intuitive? Today we dropped a first look video of what we created.PearlOS is a browser based desktop environment where the AI companion (Pearl) talks to you to open apps, manage windows, build characters, research, takes notes, searches the web, and controls the whole UI. You just talk to her and things happen on screen.It's early. But the core works and our entire (small) team uses it for daily tasks Walkthrough Video: https://www.youtube.com/watch?v=aKO52ox0dx0 GitHub: https://github.com/NiaExperience/PearlOS/We want to build this out into “personal pearls” that everyone can have with no coding/AI experi
Mar 19, 2026 3:49 PM
bennypowers.dev
Show HN: Backlit – Lit Web Component SSR for Drupal via Go and WASM, No Node.jshttps://bennypowers.dev/posts/drupal-lit-ssr-wasm/Two years ago I posted about server-rendering Lit web components in Drupal using a Node.js sidecar container. It worked but felt like a proof of concept -- Docker, inter-container networking, an extra process to babysit.Backlit replaces all of that with two lines: composer require bennypowers/backlit drush en backlit Composer downloads a pre-compiled Go binary for your platform. The binary embeds a WASM module (QuickJS + @lit-labs/ssr, compiled via Javy). Drupal pipes page HTML to the binary's stdin; rendered Declarative Shadow DOM HTML comes back on stdout. The WASM instance stays warm across requests: ~350ms cold start once per PHP-FPM worker, ~0.32ms per render after that.The funny part: I didn't build this for Drupal. I built lit-ssr-wasm to add live component previews to `cem serve` -- a dev server for custom elements manifest tooling (https://github.com/bennypowers/cem). Once the WASM module existed, the Drupal integration was an afternoon's work.WASM is to backend runtimes what
Mar 19, 2026 3:15 PM