Show HN: Haystack – Review pull requests like you wrote them yourself https://ift.tt/4kgI8sb

Show HN: Haystack – Review pull requests like you wrote them yourself Hi HN! We’re Akshay and Jake. We put together a tool called Haystack to make pull requests straightforward to read. What Haystack does: -- Builds a clear narrative. Changes in Haystack aren’t just arranged as unordered diffs. Instead, they unfold in a logical order, each paired with an explanation in plain, precise language -- Focuses attention where it counts. Routine plumbing and refactors are put into skimmable sections so you can spend your time on design and correctness -- Provides full cross-file context. Every new or changed function/variable is traced across the codebase, showing how it’s used beyond the immediate diff Here’s a quick demo: https://youtu.be/w5Lq5wBUS-I If you’d like to give it a spin, head over to haystackeditor.com/review! We set up some demo PRs that you should be able to understand and review even if you’ve never seen the repos before! We used to work at big companies, where reviewing non-trivial pull requests felt like reading a book with its pages out of order. We would jump and scroll between files, trying to piece together the author’s intent before we could even start reviewing. And, as authors, we would spend time to restructure our own commits just to make them readable. AI has made this even trickier. Today it’s not uncommon for a pull request to contain code the author doesn’t fully understand themselves! So, we built Haystack to help reviewers spend less time untangling code and more time giving meaningful feedback. We would love to hear about whether it gets the job done for you! How we got here: Haystack began as (yet another) VS Code fork where we experimented with visualizing code changes on a canvas. At first, it was a neat way to show how pieces of code worked together. But customers started laying out their entire codebase just to make sense of it. That’s when we realized the deeper problem: understanding a codebase is hard, and engineers need better ways to quickly understand unfamiliar code. As we kept building, another insight emerged: with AI woven into workflows, engineers don’t always need to master every corner of a codebase to ship features. But in code review, deep and continuous context still matters, especially to separate what’s important to review from plumbing and follow-on changes. So we pivoted. We took what we’d learned and worked closely with engineers to refine the idea. We started with simple code analysis (using language servers, tree-sitter, etc.) to show how changes relate. Then we added AI to explain and organize those changes and to trace how data moves through a pull request. Finally, we fused the two by empowering AI agents to use static analyses. Step by step, that became the Haystack we’re showing today. We’d love to hear your thoughts, feedback, or suggestions! https://ift.tt/cRG1udz September 10, 2025 at 11:51PM

Show HN: WorldView – Compare how different countries report the same news https://ift.tt/2EGTkX1

Show HN: WorldView – Compare how different countries report the same news https://worldview.up.railway.app/ September 10, 2025 at 11:47PM

Show HN: Real time visual saliency detection https://ift.tt/3wzaM2k

Show HN: Real time visual saliency detection I've just made public a library to perform real time visual saliency detection on videos (but static images are also supported). This started a couple months ago when, while working on another project, I ended up side-tracking and overkilling as usual. I'm pretty happy with the result and I think it could prove to be a useful piece of software. It should work on both Linux and macOS, but I'm yet to test Linux cause I only have a mac at hand. Windows may be doable through WSL. GitHub: https://ift.tt/tdPl4IJ Showcase: https://big-nacho.github.io/dosage-docs/showcase.html https://ift.tt/tdPl4IJ September 10, 2025 at 01:36AM

Show HN: An Open Source XR(AR/VR) Operating System https://ift.tt/Fys1YIE

Show HN: An Open Source XR(AR/VR) Operating System We're two college students building an XR(AR/VR) native Operating System with a custom kernel. We're also Open Source so feel free to check our GitHub Repository- https://ift.tt/ueK9TZs . The journey hasn't exactly been easy, we've been criticized by a lot saying that whatever we're doing is impractical and that we're too ambitious. Regardless, we've been committed to reach our goal. Here to answer all questions and doubts. Answering one question beforehand because we know someone is going to ask it - Q: Why use your own kernel/ Why don't you use Linux/ Why are you trying to reinvent the wheel? A: Using our own kernel helps us get rid of the baggage of legacy codes, bring the most optimal performance on our target hardware (XR/AR/VR) and achieve more efficiency than what we would've achieved on an existing kernel. We're not trying to reinvent the wheel, but just building Formula One racing tyres for it. https://ift.tt/aXHlW0S September 7, 2025 at 04:39PM

Show HN: Superagents – connect spreadsheets to any database, API or MCP server https://ift.tt/BNoMYms

Show HN: Superagents – connect spreadsheets to any database, API or MCP server Hi HN, I’m Eoin, founder of Sourcetable ( https://sourcetable.com ). Today, we’re launching Superagents. You can now connect your spreadsheet to any database, API or MCP server on the Internet. All of that data is available inside your spreadsheet, and you can use AI to analyze it and build models, reports and visualizations. The reason I started the company is because I spent 10 years at startups across engineering and operations roles and realized that Excel and Sheets weren't architected for the modern information environment. This creates a tremendous amount of nuisance and busywork cobbling together SaaS tools, reporting suites, and the misery of endless coordination meetings to make it all happen. (Boo meetings!) Spreadsheets aren’t just a business application: they’re the original thinking tool. The quality of these tools has a downstream impact on analytical thinking and creativity writ large, so this is a problem worth solving. Fast forward to today, we’re a 6 person team taking on Excel, Sheets and ChatGPT, so we’re excited to hear what you think! Who are Superagents for? Analysts, operators, and anyone doing data-centric work in spreadsheets. We see a tonne of finance people, of course, but also students, researchers and mom & pop shops. Sourcetable's superagents democratize data access and analysis, which is nice because our company’s mission is to make data accessible to everyone. Why “Superagents”? Because they can plan and orchestrate other task-specific agents to complete your work for you. We have a lot of different AI tools and agents inside Sourcetable, but there’s a whole lot more on the Agentic Web. Superagents are like the conductor that coordinates them all and calls on them when needed. Also, it’s a fun feature name (thanks, Alyssa!) If you remember the linked-data dream of the semantic web movement, that future is now: all of your business data is available and connected in Sourcetable. How does it work? Sourcetable is running a python virtual machine under the hood. Everything is sandboxed, and there are hundreds of AI tools and libraries our AI can access. Superagents are also doing code-gen on the fly to solve problems. The closest system we have found is Replit’s sandboxed operating systems. Beyond that Mixtral, ChatGPT and Anthropic offer some limited data connectivity features, except these AI chat services lack the storage, compute, and code execution that Sourcetable and Replit provide. This is all very new. How is this different to your previous data connectors, etc? We started out using ETL services to sync data and provide a GUI-driven PowerBI like experience in your spreadsheet. This was useful for people who knew SQL and how to write joins to combine fragmented data, but for everyone else (read: practically everyone), this solution just didn’t provide the frictionless, self-serve experience that we wanted. Our choices were to switch the GTM motion or change the product, so we shelved that reporting suite and focused on our AI spreadsheet and waited for models to catch up with our ambitions. Now that they have, we’re re-launching Sourcetable with our original goal in mind: building a spreadsheet-based operating system for the Agent Web, with fully networked data access for everyone on your team. AI is the great UX enabler. Caveats: * We heavily use Postgres, Google Analytics, Stripe and Google Search Console with Superagents. * We haven’t tested every endpoint on the Internet. We find that mainstream, well documented applications work best. * Yes, you can write data back to 3rd party applications and databases. We generally advise against this unless you understand the risks involved in giving AI write-access to your data. Bonus round: * All data connectors added during this launch week are FREE. (Regular AI messaging limits still apply.) Product Feedback? eoin@sourcetable.com https://ift.tt/TGw9L0E September 10, 2025 at 12:25AM

Show HN: C++ Compiler Support Page https://ift.tt/54uoZBk

Show HN: C++ Compiler Support Page Hi HN, I have created a webpage that displays all C++ features since C++20 in a simple, searchable table. It is intended to serve as a quick reference for C++ developers, whether as support for cross-platform development or simply to track the current support status out of curiosity. I created it as a simpler, more structured, and more up-to-date alternative to the cppreference compiler support site. Please note that the page intentionally does not list LWG and CWG papers. This might change as I am continually updating the site and trying out new ideas. Questions, feedback and suggestions are appreciated, either here or in the form of GitHub issues. https://cppstat.dev September 8, 2025 at 12:42PM

Show HN: I made a simple ASCII-art analog clock in Emacs https://ift.tt/7wfKHg3

Show HN: I made a simple ASCII-art analog clock in Emacs Just a toy, showing how easy it is to leverage built-in Emacs features (most notably Artist mode, which provides a set of functions for creating ASCII-art vector graphics) and things like trigonometric functions and timers to create something nice. A short blog post mentioning some background (and showing a screenshot): https://ift.tt/LZfhN8W . https://ift.tt/flizqFv September 8, 2025 at 11:48PM

Show HN: Vizzly – Visual testing platform with built-in review workflows https://ift.tt/PRWud59

Show HN: Vizzly – Visual testing platform with built-in review workflows Hey HN! I’m Robert. I worked on [Percy’s SDKs/support from 2018–2022.]( https://ift.tt/MFfw14Y ) If you ever reached out to support or opened an issue, I probably helped you. Hopefully it was positive :) After a few years away, visual testing still felt stuck, so I'm building Vizzly. The problem: Design handoffs kinda suck. Designers make beautiful specs, devs implement them, then everyone realizes the details got lost somewhere. Current visual testing tools catch broken CSS but miss the real issue - making sure what ships actually matches what was designed, functionally (like really in the browser/OS/etc). What it is - Visual testing + review where you send actual screenshots (not DOM re-renders). Can be _any_ image to review (PDFS!) - Collaboration built-in: reviewer assignment, approvals, @mentions, screenshot-level threads. - Baselines: automatic (Git-aware), manual (not Git-based), or hybrid. - Team-based pricing; generous free plan for OSS; on-prem available. What’s different - Capture-first: use the pixels your app produced (no “but it doesn’t look like that on my machine/CI”). - Local TDD + CI parity: run locally with instant feedback; same flow in CI. - Custom properties to filter/slice reviews (component, viewport, theme, etc). Try it quickly (Playwright example) ``` npm i -D @vizzly-testing/cli export VIZZLY_TOKEN=your-token # in your tests: import { vizzlyScreenshot } from '@vizzly-testing/cli'; let img = await page.screenshot({ fullPage: true }); await vizzlyScreenshot('homepage-layout', img); ``` I would love feedback on everything! Rough edges you hit using the product/sdk, baseline expectations across branches, what you need for design/dev review to feel “done”, etc. Features like root cause analysis, an MCP, and more collab features are coming. But it's just me building :p I'm a big fan of OSS, so the OSS plan is pretty generous (10 seats + 10 review seats (20 total), unlimited public projects, 75GB, 6 concurrent builds). If it's not generous enough for teams, I'm willing to up it! This is my first time launching anything like this, I'm super keen on getting feedback and working any support or suggestions folks have. If anyone knew me from my support at Percy, I _really_ enjoy those conversations and opportunities to ship a fix or feature at the end of a chat. If Vizzly isn't it for your team, I wanna know why and what I can do to help you. Backstory + screenshots from my intro blog post: https://ift.tt/yY2N0Kl https://vizzly.dev September 7, 2025 at 09:14PM

Show HN: Psq – CLI for Postgres Monitoring https://ift.tt/ELTXqCK

Show HN: Psq – CLI for Postgres Monitoring https://ift.tt/Wd0cBQI September 8, 2025 at 12:11AM

Show HN: The World After 3, 5, 10, 25, 50, and 100 Years Ft. AI https://ift.tt/P9XiFWA

Show HN: The World After 3, 5, 10, 25, 50, and 100 Years Ft. AI AI is arguably the greatest invention in modern human history. Humanity has always evolved in hockeystick curves, each major discovery unlocking an entirely new trajectory of progress. But what does this mean for us, Humans ? dive in for more info here⬇ https://ift.tt/fFGPgmc... https://ift.tt/vRcSQB0 September 8, 2025 at 12:24AM

Show HN: rm-safely – A shell alias that moves files to trash instead of deleting https://ift.tt/UwgfhtP

Show HN: rm-safely – A shell alias that moves files to trash instead of deleting I made rm-safely, a simple shell wrapper that moves files to trash instead of permanently deleting them. It prevents accidental deletions from autocomplete mishaps or hasty rm -rf commands. Should work as a drop-in replacement for rm but safer. Would appreciate any feedback! https://ift.tt/dHwK3jQ September 4, 2025 at 12:38PM

Show HN: Evalyze – AI investor matching from your pitch deck (feedback welcome) https://ift.tt/nvU39ME

Show HN: Evalyze – AI investor matching from your pitch deck (feedback welcome) I used to work in VC and watched good teams lose months chasing the wrong investors. I’m building Evalyze to make the unglamorous parts faster and more precise. After sign-up (email only, no card) you can: - upload a deck or paste your site - get a ranked list of relevant VCs/angels with a short “why” for each What’s different: instead of dumping a big list, we try to explain why an investor fits based on stage, sector, check size, and portfolio patterns. It’s far from perfect and we want blunt feedback before opening wider. Limits to know: - newer funds and emerging managers can be underrepresented - geo nuances are still rough - matching can over-weight buzzwords if the deck is vague I’d love critique on the ranking logic, signals you’d add/remove, and any privacy concerns. If you don’t want to upload a deck, there’s a sample you can use to see the flow. I’ll be here replying and shipping fixes as comments come in. https://ift.tt/wPh27qv September 6, 2025 at 11:10PM

Show HN: 60-Second Linux Analysis, Supercharged with Nix and LLMs https://ift.tt/7rZsjBp

Show HN: 60-Second Linux Analysis, Supercharged with Nix and LLMs Hello HN, I'm sharing a little open-source utility I wrote recently. I'm a huge fan of Brendan Gregg's "BPF Performance Tools" book. However, every time I SSH into a fresh server, most of the diagnostic tools aren't installed there and installing them can be really annoying. I decided to use Nix package manager and LLMs to make this process straightforward. My utility first downloads a "toolbox" of Linux utilities (built with Nix), runs Brendan Gregg's famous "60-second Linux analysis" playbook and then summarizes the results with an LLM. So "60-second Linux analysis" now becomes a single one-line command and actually takes less than 60 seconds! The utility can execute all commands in parallel and the LLM can analyze them faster than a human would. I have a few ideas for the future, for example implementing more powerful playbooks - thanks to Nix I can easily bundle all tools I need and LLMs have no trouble analyzing outputs of tens of commands. I'd love to get your feedback and hear any ideas you have. Thanks for checking it out. You can launch the utility with this command: $ curl -fsSL https://ift.tt/mjD42UJ | sh https://ift.tt/JoSyuzr September 6, 2025 at 09:23PM

Show HN: Find the cheapest protein per gram across 3000 powders https://ift.tt/2dnzjZ6

Show HN: Find the cheapest protein per gram across 3000 powders I tracked protein powder prices in a spreadsheet for years and found that identical protein content can have wild price differences. I built PricePerProtein to automate it. It pulls real-time Amazon data (Keepa API) and uses Gemini 2.5 Flash to extract nutrition facts from product images/descriptions. Calculates actual protein per dollar, not just package price. Technical: FastAPI + Celery backend, Next.js frontend with virtual scrolling to handle 3000+ products. Deployed on a VPS (migrated from GCP - much simpler). The AI handles everything from blurry nutrition labels to understanding flavor categories. No signup, no ads, no affiliate links. Updates hourly. https://ift.tt/yaM8OSA September 7, 2025 at 12:18AM

Show HN: Writing Arabic in English https://ift.tt/iKBQ8ft

Show HN: Writing Arabic in English A phonetic Arabic keyboard I created maps English letters to Arabic sounds, covering emphatic letters, hamza, and diacritics—making it easier for learners and casual users to type Arabic. https://ift.tt/lRAsW3H September 3, 2025 at 07:34PM

Show HN: I built a public and open llms.txt endpoint for every domain https://ift.tt/u1Dgohy

Show HN: I built a public and open llms.txt endpoint for every domain And yes, I know, literally no AI uses llms.txt right now. But hear me out: if you want it just in case, or if you would like to add your sites to some llms.txt directories, you can use this endpoint. That way, you do not need to keep updating your own llms.txt, especially as I improve the API. Here is how it works: Enter any domain: https://get.llms.page/{example.com}/llms.txt The API will parse your homepage (if allowed in robots). Using internal links, descriptions, and other metadata, it will generate an Markdown llms.txt file. It does not rely on AI, because I want it to be fast and free. The API is open, free, runs on a CDN, and is powered by Cloudflare Workers for speed. I plan to open source the no-AI llms.txt generator later, since there is still a lot to improve. If you want to try it out or see some usage examples, visit: https://llms.page Let me know what you think! https://llms.page/ September 6, 2025 at 01:45AM

Show HN: Open-sourcing our text-to-CAD app https://ift.tt/UCtZREJ

Show HN: Open-sourcing our text-to-CAD app Hey HN! I'm Zach from Adam ( https://adam.new/ ). We’re building an AI co-pilot for mechanical CAD software. As part of our broader research, we built a browser-based Text-to-CAD app ( https://ift.tt/WZChiNm ) and are now open sourcing it. This is a React SPA with a Supabase backend. What it does: * Generates parametric 3D models from natural language descriptions, with support for both text prompts and image references * Outputs OpenSCAD code with automatically extracted parameters that surface as interactive sliders for instant dimension tweaking * Exports as .STL or .SCAD Under the hood: * Separate agents for conversation and code generation; simple parameter tweaks bypass AI entirely using deterministic regex-based updates * Runs fully in-browser by compiling OpenSCAD to WebAssembly and integrating Three.js with React Three Fiber for 3D rendering * Supports BOSL, BOSL2, MCAD libraries and custom font support (Geist) for text in models We’ve seen many developers trying to replicate this kind of functionality, so we’re releasing this to give the community a solid foundation to build on. Future improvements: * Expand geometry support - Move beyond CSG primitives to support curved surfaces, fillets, lofts, and constraint-driven modeling through CadQuery/Build123D * Better spatial context - UI for face/edge selection and viewport image integration to give LLMs spatial understanding * Enhanced capabilities - RAG on documentation and integration with more OpenSCAD libraries for features like proper threading You can clone the repo and run it locally! Contributions are welcome, and we’ll keep merging PRs as they come in. https://ift.tt/az6W0q9 September 5, 2025 at 10:39PM

Show HN: A small browser game (PC only) built with Phaser 3 https://ift.tt/fMbvrm2

Show HN: A small browser game (PC only) built with Phaser 3 Hi HN! This is my first game — something I’ve always wanted to do. It’s a small browser game built with Phaser 3, React, and the phaser template ( https://ift.tt/s6r3f1o ). I made it in 2 days (like 8 hours in total real time) using gemini-cli. About 90% of the code was generated with AI, but I learned a lot by making fine tweaks. It only works on PC since it’s a typical WASD + R (reload) shooter. I’d love feedback on: - Gameplay (is it fun, too hard?) - Ideas for new features Thanks in advance! ps: I used cubes as a prototype, but now I kind of like them. Should I keep them or implement proper sprites? https://cubic-zombies.pages.dev/ September 5, 2025 at 02:44AM

Show HN: Comfyfile - Secure, Anonymous File Sharing with Auto-Expiry No Account https://ift.tt/Ukw1d0e

Show HN: Comfyfile - Secure, Anonymous File Sharing with Auto-Expiry No Account https://comfyfile.com September 4, 2025 at 09:04PM

Show HN: Quicknote.zip (Daily Micro Scratchpad) https://ift.tt/u3ZGT6D

Show HN: Quicknote.zip (Daily Micro Scratchpad) I used to use https://doc.new when I needed to write quick scratchpad notes, but it takes like two seconds for Google Docs to be interactable, and ends up polluting my Drive with a bunch of "Untitled Docs". Lately I've used a bookmarklet that opens a fullpage contenteditable div which is instantaneous and worked for my needs. But I wanted persistence when I accidentally close the tab, and data-urls can't use localstorage, so I spun up quicknote.zip. It loads in the blink of an eye, works offline, and stores each day to localstorage. That's all it does, take it or leave it. https://quicknote.zip September 4, 2025 at 11:40PM