8,948 results for now

github.com
Show HN: Voro – An attention manager for agentic codinghttps://github.com/ClachDev/VoroA few months ago I was getting frustrated trying to use Claude Code effectively. I was finding that for small tasks I would spend a lot of time waiting for it to finish "thinking" and I would end up browsing here...To make a better use of my time, and ideally make better use of the limits, I developed Voro to help me know what I should be focusing on.Fundamentally it's a task tracker tailored for local agentic coding. Each task is a plan that gets a priority and works its way through a set of states that are either for "humans" or "agents". The main "cockpit" view lists a few different options of what I should be looking at. The bottom has a list of what your agents are working on.Agents are given a small description and generate a task. That can then be reviewed and accepted if it's good, or iterated on by refining. Once accepted it can be dispatched to an agent for implementation. Once finished it comes back for review which can again then be iterated on. The crucial thing is that ma
Aug 21, 2026 10:07 AM
wired.com
The Super El Niño Won’t Fix the West’s Water Crisishttps://www.wired.com/story/the-super-el-nino-wont-fix-the-wests-water-crisis/“People have talked about the Colorado River crisis for decades—we’re now in it,” says one expert. “Everyone you talk to is like, ‘Shit, man.’”
Aug 21, 2026 10:00 AM
news.ycombinator.com
Ask HN: Has Microsoft forgotten how to test software?https://news.ycombinator.com/item?id=49383954This is a serious question.I've an older Mac with some critical apps that runs a slightly obsolete macOS. This OS version is no longer supported by Edge and Chrome 151+.This did not stop Edge from auto-updating to a known incompatible version and bricking itself.One minute I had the browser open; the next, the icon was overlayed with the Ghostbusters symbol and could do nothing but display a dialog explaining that my OS is unsupported.A fluke once perhaps, but this is the third time this has happened to me with various MS apps bricking themselves.I'm officially done with Microsoft products. This isn't some annoying bug; this is suddenly and abruptly locking me out of my own computer.
Aug 21, 2026 4:55 AM
news.ycombinator.com
Receive calls only from saved contactshttps://news.ycombinator.com/item?id=49383219Hello folks, I built an app to block calls from non saved contacts, it works for Android.I live in Brazil, and at least here the situation with spam and bot calls is pretty bad, some days you can get calls from unknown numbers constantly.If someone wants to talk to me, they can usually reach me asynchronously (email or whatsapp) and I answer when I can/want. If it’s urgent, the people I care about enough to receive an immediate phone call from are generally already saved in my contacts.So Fone1 does one thing: when enabled, it blocks calls from numbers that aren’t saved in your contacts.I know this usage pattern definitely isn’t for everyone, but if you feel similarly, I’d love for you to try it and share feedback.It’s free and has no ads.If you are an Android user and wants to test, just send me your email so I can add you to list of testers.
Aug 21, 2026 3:01 AM
news.ycombinator.com
Ask HN: Is building military hardware and healthcare the AI-proof?https://news.ycombinator.com/item?id=49382011As developing software as a skill becomes like flipping burgers in a fast food company, is building hardware for evergreen industries like military and healthcare the only escape for actual engineers, not prompters?Sure, it would be a lot more expensive to bring a product to market and sell, and more likely for a product to fail because of the long sale period. But at least you have a natural barrier then. You don't see some solo dev trying to compete with Apple in hardware, whereas in software nowadays there are billions of people who are your competition.
Aug 21, 2026 12:12 AM
basecompute.co
Show HN: Zero () friction local AI for Machttps://www.basecompute.co/localSuper excited to launch our new app Local today. What we’ve learned at Base Compute over the last months is that running AI directly on your laptop or workstation gives you maximum privacy and it’s free, but it’s also a massive headache to configure. So we’ve decided what matters is making the experience completely frictionless for users.Local analyses the hardware of your laptop, optimises the AI for it, and recommends the best models for your specific device.It let’s you do what you’re doing with cloud AI already, just for free and on your own machine: Chatting with PDF’s, Recording and summarising meetings, running coding agents...If you’re using Local in your office with colleagues, you can run it in “Office Mode”. The strongest computer in your office runs the AI and everyone can connect to it with their laptop. The data never leaves the office.It’s available for download on our website today, please try it out and let us know what you think!
Aug 20, 2026 10:13 PM
youtilitics.com
Show HN: Youtilitics, energy usage analytics from utility data, no hardwarehttps://youtilitics.com/I posted an early version of this 2 years ago. Rebuilt most of it since then, so posting again with what's actually different.Youtilitics still pulls interval data your utility already collects (Green Button/ESPI — 15-60 min whole-house kWh) and turns it into usage analytics. The analytics stack is different: baseline anomaly detection (median/MAD by time-of-day bucket, not a flat threshold), overnight step-change alerts, usage heatmaps, cycle-over-cycle comparisons, etc. It is still based on a model trained on human-labelled data: used to differentiate baseline (eg vampire load) vs EV vs HVAC vs the rest. Inference runs with a python script when new data is received from the utility.I also changed the mobile app. It is still mainly used to store the user credentials without us having them (they never leave the device), for security/privacy reasons. This is used to scrape usage data from the utility dashboard, when they don't support Green Button. However now it also runs an embedded P
Aug 20, 2026 10:13 PM
praxos.ai
Show HN: Praxos – team messaging with built-in memoryhttps://www.praxos.ai/Hey HN! What does a regular workflow at a startup look like?Probably like this:> One founder talks to customers> Another founder builds> A customer asks for something. The first founder explains it to the second. The second founder explains it to Claude> Two weeks any one of the founders asks one of the following questions: \* Why are we building it this way? \* What happened here? \* What did we promise ACME we’d do? \* How far along are we on X? > Then you scream on the inside because you have to figure out what happened...Before agents, the person who did one part of the work usually held a mental map. They could share the context with other people. Increasingly, they can’t.Small teams can run far more work in parallel than ever before thanks to AI, but someone now has to connect all that context manually. The reasoning behind one product change might now be scattered across a customer call, two emails, a Slack thread, and four Claude conversations.Praxos is a team messaging app.It
Aug 20, 2026 8:55 PM
pingchange.com
Show HN: PingChange – monitor web pages and OpenAPI specs for meaningful changeshttps://pingchange.comI built PingChange after watching the same thing happen over and over: a competitor prices change, a company policy is updated or important web content changes but there is often no reliable way to track it and know when it happens.The tool monitors web pages and API on a schedule and alerts your team when something meaningful changes. The keyword here is meaningful. Most monitoring tools trigger alerts for everything (ads, timestamps, cookie banners etc) and could be noisy or spam in your inbox. With PingChange, you can give it an instruction e.g "alert me when pricing changes" and it analyzes the page content and only notifies you when the changes matches what you're looking for on the page.Here are the main features:- Visual region selection / HTML element selector for tracking specific location on the page. - Changed diff viewer and screenshot of the pages. - Geolocation support - Monitor URL's from specific location to detect content from different geographies. - Anti-bot resoluti
Aug 20, 2026 8:51 PM
news.ycombinator.com
Qwen3.8 Fetches Weird URLshttps://news.ycombinator.com/item?id=49379079I am working on opencode with the last qwen3.8-27B from unsloth (with dynamic 3.0 quants).while checking documentation for a rust project, instead of fetching the url it had listed in its own reasoning, it tried fetching urls like:https://routify-file-proxy-sg.oss-ap-southeast-1.aliyuncs.com/proxy_temp_file/production/2026-08-20/trace_0baf8c2b17874702866132490e0b56/requestId_97422c0041074091a68367881782d7b7/7c0791038d2f58b3a52c0f07714d3627.html?Expires=1818574288&OSSAccessKeyId=LTAI5t…QVZr&Signature=96y%2B…w%3Dthis happened in the same session, different batches, 16 times in total.It convinced itself there was some rewriting proxy in the webfetch tool.`Expires`, `trace` and `requestId` were consistent for a few requests, but they do change.`AccessKeyId` was always the same, "LTAI5t…QVZr"I did not add the `…` in the signature and access key.Probably training links, but that `Expires` number seems very high...in unix timestamp:1818574288 -- Aug 20271787251180 -- nowJust leaving this here
Aug 20, 2026 7:32 PM
danielvaughn.dev
Show HN: Huzzah – a novel approach to coding with AIhttps://www.danielvaughn.dev/posts/huzzah/Hello everyone. I've been working on this experimental editor called Huzzah.I've been working almost exclusively with coding agents since January of this year, and over the past few months I began to feel utterly exhausted by them. They're great, but I'm finding it more and more tedious to write full sentences for every change I want. Not only that, but it seems there's a complexity limit for codebases - beyond a certain point the agent begins confusing itself.I'd like to go back to writing code, but I don't want to go all the way back to fully manual coding. So I've come up with this interaction paradigm where you: 1. write pseudocode in whatever way makes the most sense to you 2. on save, the editor synchronizes your work to real source code 3. the pseudocode is persisted alongside the generated code, making your prompt effectively a stored record of intent. It may not work for every use case, but in my initial playthroughs I've found it very enjoyable.Right now it's just a proof of
Aug 20, 2026 7:05 PM
news.ycombinator.com
InstantDB Is Shutting Downhttps://news.ycombinator.com/item?id=49378706Nothing on their website - but I got an email.Hey there, you’re one of our active users and we wanted to let you know we’re sunsetting Instant.In more detail:New signups will be closed soon. We recommend migrating your apps off of Instant Cloud within the next 12 months. Any subscriptions that were started after July 31st, 2026 will be fully refunded. On August 31st, 2027, all cloud apps will shut down. Backups will stay available for 12 more months, until August 31st, 2028. We’ve written a guide to help you self-host Instant and migrate your apps too. All of Instant is open source, and we’ve worked to make the migration process as seamless as possible. The idea behind Instant started in 2021. We shared a blog post about a database in the browser. It was only an idea, but the more apps we built, the more we saw the need for it. We joined YC in 2022, built a prototype, and shared a version zero.Turning a prototype into something production-grade was a long journey. The work was hard, bu
Aug 20, 2026 6:59 PM
news.ycombinator.com
Ask HN: How do you review and validate LLM generated code?https://news.ycombinator.com/item?id=49378314It's become very fast to generate code nowadays, but how do you review and test all of it?Every time you ask a model to review some piece of code, it comes back with different "findings". So, you still have to filter through the noise of their findings to get to actual defects.And if you try to do the full review manually, then it might take more time to just understand the code than writing it yourself from the beginning.As for testing, you can ask the LLM to write the tests, but you can't be sure if the tests actually assert the expected product flow or just write pass/fail tests for a piece of logic.I would like to know if you have any workflows, strategies, or tricks.
Aug 20, 2026 6:26 PM
news.ycombinator.com
Are you good at AI, or just using it?https://news.ycombinator.com/item?id=49378057We’re working on a ladder for individual AI proficiency and would love feedback on both the levels and the definitions.L0 New: brand new to AI, or has not yet used it.L1 Chat: simple prompt-and-response use. Work is serial: ask, wait for an answer, then ask again.L2 Contextual Work: gives AI relevant documents, data, or workspace context so it can work within the actual artifact and produce a more useful result.L3 Orchestrate: coordinates multiple agents or AI roles across independent workstreams, with work that may review, challenge, compare, or build on other work. This is not just for engineering.L4 Automate: creates workflows that are triggered by business events and run without someone sitting at a laptop directing each step.L5 Loop: feeds the output of those workflows back into shared knowledge or a company brain, so future workflows improve over time.A few things I’d love your perspective on:Are these the right levels? Are any of the names unclear or overlapping? What observable
Aug 20, 2026 6:04 PM
foundera.app
Show HN: AI-Native Accelerator that helps builders make something people wanthttps://foundera.app/Most people are not validating their ideas before start building it! I am a 5th time tech founder, gaming included and I built & managed startup accelerator programs. I know lot of things about early stage founders and their pains as I have been there. That is why I am building Foundera. It’s an AI-Native Accelerator for tech builders. Today the problem is; most of the AI coding tools opens with a chat and asking “type your idea and start building”. This is wrong! I want to reduce the time and money spent on ideas that nobody asked for. Try it at https://foundera.app and give me your feedback so that I can help more people.
Aug 20, 2026 5:16 PM
news.ycombinator.com
Ask HN: How to engage users in an open source project and when to stop?https://news.ycombinator.com/item?id=49377336I have been working on an OS project for years now, on and off, taking it back up for the last few months and giving it a good boost lately with AI. It has a super niche user base in quantitative finance.I really enjoy working on it, and although I get contacted every now and then by people who are using it or are interested in using it, and those are really interesting users, I always think I am wasting my time. It has been 8 years now since I started.I guess my questions are: is it worth it? And I know this question has two clear paths, the personal and the practical one, but I would like to hear from both. How can I know when to stop? If my audience is maybe in the order of two thousand potential users worldwide, how do I validate the idea? I have posted on Reddit and HN multiple times, tried LinkedIn, and that leads me to a handful of super interesting contacts, but I don't know if it is enough to keep going.
Aug 20, 2026 5:09 PM
tablecanon.app
Show HN: Building Table Canon, an AI Campaign Memory Engine for TTRPGshttps://tablecanon.app/Hey HN! I built Table Canon to solve a problem my playgroup kept running into: 3-4 hour tabletop gaming sessions leave behind massive audio recordings, but standard meeting note-takers treat every session as an isolated island, butcher fantasy terms, and don't know who is speaking.I wanted an engine that tracks long-term state across months of games, so I built a pipeline to extract entity updates, open quest hooks, and character promises across sessions.The Tech Stack:* Transcription: whisper-large-v3-turbo * Diarization: pyannote for speaker embeddings & voice profile matching Extraction & Memory: OpenAI API with Structured Outputs (JSON Schema enforcement for state updates) * TTS & Audio Recaps: Kokoro / Chatterbox Turbo Music Generation: ACE-Step-v1.5-XL-Turbo for rendering session summaries into lyrics/balladsA Few Engineering Lessons & Challenges:* State Delta Extraction vs. Context Explosions: Feeding 20 prior session transcripts into context windows quickly becomes cost-prohibi
Aug 20, 2026 4:40 PM
github.com
Show HN: Pond – lossless archive for agent sessions in your own S3https://github.com/tenequm/pondI wanted all of my sessions preserved in one S3 storage dump I could access from anywhere. I couldn't find anything that could store sessions from multiple machines in one remote S3 bucket without running a database service anywhere, make them available as an MCP for my agent, and efficiently search through them. pond was born out of this, and out of my necessity to stop being locked to my laptop. I'm convinced that the sessions we generate with agents are one of the most precious assets we have and it still feels so odd to me that we throw them away like we couldn't care less.With pond you just point it at an S3 bucket or a local directory. Under the hood is in-process Lance. It also supports safe concurrent writes, so I just collect all my sessions from all my local machines, remote VMs and telegram agent setups in one bucket dir. Right now that is 14,861 sessions, 2.83M messages, 10.6 GiB, from 8 tools (Claude Code, Codex, opencode, pi, Claude Desktop, oh-my-pi and a few more).The o
Aug 20, 2026 4:04 PM