Ask HN: How do I reliably eval my AI modelsI created an AI agent and have been considering using Braintrust or Arize - but where can I source experts who actually can make goldsets for me too? Is it worth sourcing them?
Show HN: Effort Router: Intelligent /effort selection per Claude turnDynamic effort selection, while possible in some other tools like OpenCode, does not exist in Claude Code.Having dynamic effort set per conversation turn means:- Less tokens burned on easy queries (e.g. something low /effort can do goes to xhigh because it's the convo default)- You avoid mistakes: Prompts requiring high effort (like a refactor or security threat modeling) go to high or xhigh instead of medium to reduce likelihood of mistakes and reduce wasted tokens.I created this out of a personal need and it's been game-changing. My Claude sessions are way more productive with less back/forth.Give it a try and feel free to contribute!
Show HN: VoxThermic – A macOS journaling app that uses Apple's Foundation ModelsTL;DR version: I made a macOS journaling app that use Foundation Models to analyze entries for mood + written themes, and also has voice transcription, because I couldn't find any other journaling app that does those things.OK, now for the longer, I-rambled-too-much version:For the past few weeks I've been working on a macOS journaling app called VoxThermic.The idea for it came from my own journaling habits, where my mood tends to shift a lot depending on whatever I'm dealing with. I've tried some journaling apps, but I've never found one that can analyze mood or see how my mood changes depending on what I'm writing about.So I just figured I'd build it myself. I've been experimenting a lot with Apple's Foundation Models and Natural Language frameworks, and those all work on-device and offline (which is good because there's no way I'm sending journal entries to some AI company's data centers!)But there's one problem with using Foundation Models: they're tiny. About 3 billion parameters,
Show HN: Sidereon: GNSS and Astrodynamics (Python, Rust, JavaScript, C, Elixir)I’ve been building Sidereon, an MIT-licensed GNSS positioning and astrodynamics engine.The submission is a browser demo running the Rust core client-side through WASM. Source for this:https://github.com/neilberkman/sidereonThe idea is one engine for both the receiver and spacecraft sides: SPP/RTK/PPP, RINEX/RTCM/NTRIP, Galileo HAS/SSR, GNSS/INS fusion, SGP4 and numerical propagation, orbit determination, conjunction screening, pass prediction, reference frames, and time systems. Python, Rust, JavaScript/WASM, C, and Elixir all use the same core.Several features came directly from open issues in widely used GNSS and astrodynamics libraries: multi-constellation ARAIM, Galileo state-space corrections, station-displacement corrections, sidereal multipath filtering, and track cleanup without an IMU.Tests compare results with RTKLIB, the 33-satellite Vallado/CelesTrak SGP4 verification corpus, pinned Skyfield frame vectors, Karney’s test set, and the published WG-C ARAIM example. The Vallado
Lakers reportedly checked in on Mavericks' $88M lockdown defender as Jonathan Kuminga plan BAs the Los Angeles Lakers continue to build their roster, their attempt at landing Jonathan Kuminga has seen them adopt a new approach. Kuminga has been a rumoured sign-and-trade target for the Lakers ...
Show HN: Read the Tape – Wordle for daytrading, five blind S&P 500 charts a dayRead the Tape gives players the same 5 S&P500 stock charts per day to predict. You select low, medium or high confidence and then call the chart UP or DOWN. It's a 1d chart which then resolves over 5 days. Alpha is scored against the Monkey Index, a basket of 11 random coin flips at low confidence which provides a tangible win/lose condition.We're two weeks in and some interesting data is being kicked up. Players like to call tops even though stonks go up- 60% of the 70 charts so far resolve higher, players' down calls have only been right 31% of the time. There's a full stats dive at https://readthetape.cc/notes/tape-report-1Your feedback and thoughts is most welcome.
Patagonia Just Slashed the Price of One of Its Most Iconic Outdoor EssentialsPatagonia’s Capilene line has been a cold-weather staple for decades, and the midweight tier sits squarely at the sweet spot of the range — warm enough for sub-freezing days, breathable enough to wear ...
Conditional Value at Risk (CVaR): Expert Guide, Uses, and FormulaJames Chen, CMT is an expert trader, investment adviser, and global market strategist. Gordon Scott has been an active investor and technical analyst for 20+ years. He is a Chartered Market Technician ...
Coding Skills Development ReportBLOG: https://juntz-g1thub.github.io/#post=coding-skills-dev REPO: https://github.com/juntz-g1thub/CodingSkills.gitBackground Recently I've been experimenting with vibe coding. I'm using OpenCode as my development tool. When developing TUI programs, AI-assisted code modification proved extremely difficult — every small bug fix required repeated manual testing, and feeding those results back to the agent was cumbersome. Especially for TUI-level feedback, I'd either take screenshots manually or type prompts. At one point, I nearly gave up on AI-assisted development.Motivated by this, I started researching MCP tools. To help agents better utilize these MCP tools, I developed a set of skills combining one or more of them. These skills have been packaged and published to npm.Released Skills 1. agc-debug — TUI Program Debugging Workflow Problem Solved When developing TUI (Terminal User Interface) programs, AI code changes require repeated manual testing. TUI-level feedback is particularly pa
Show HN: A canvas-based note taking and organizer appSo I've been working on this app for a very long time now. Started off by wanting an app that separates long form notes from short form notes visually and keeps everything organized on a canvas.The app features:-Sticky notes for quick, small notes and A4 notes for long form documents-A quick mode that uses local storage to quickly open the app and jot down a thought on the canvas -A PDF export option that exports your notes into 3 custom designed PDF styles-Visual hierarchy to replicate file structure instead of a linear fashion, where your files are notes, stacks are folders and labels are names for your foldersWould love to continue working on this app if you guys like the concept.
Show HN: My Fable 5 Project Was a Multiraft Database and Blob StoreLinked from this post is the visualizer and simulator I had Fable 5 create to demonstrate theIn a thread the other day, I asserted it was nice that code is no longer a moat. Getting Raft or Paxos consensus working is one of the most challenging tasks out there, even which etcd.io/raft and hashicorp/raft.The database took about a day for Fable 5 to create. It was updated as tons of work was done on a sandbox app running on top of it.What it delivered was a distributed unified KV and blob store. It automatically does sharding and erasure coding.You'll find something named Private Cloud Platform in the `sandbox/` directory. Will be making a video on it soon going over the ridiculous amount of features it has packed into it. There's some things not fully fleshed out like the docs and spreadsheets (no import/export or cross format support yet), but has other features like a Markdown document editor, Kanbans, and Diagrams.Most of the work since Fable 5 service was restored has been in the Pr
Fields Medal 2026: Leaked names reveal a century of math solved at onceFour days before the International Congress of Mathematicians opens in Philadelphia, the world's most prestigious mathematics prize appears to have been spoiled by a ...
Jennifer Finch, bassist for grunge rock feminists L7, dies at 59Grunge trailblazers L7 honour their "beloved bandmate, friend, and fellow troublemaker", who died just five days after revealing she had been diagnosed with aggressive brain cancer.
Remembering Nico 38 Years After the Death of the Velvet Underground SingerAdd Yahoo as a preferred source to see more of our stories on Google. In July 1988, model and singer Nico passed away aged 49 after falling from her bicycle in Spain. Nico had been a model for Coco ...
Concrete Slab-On-Ground Thickness Requirements From Modern DesignThe concrete slab-on-ground is among the most performance-critical structural elements in industrial and commercial buildings, yet it has long been designed using methods whose fundamental limitations ...
Show HN: I replaced a $120k bowling center system with $1,600 in ESP32sI might be the only SRE on Earth with his own bowling center. It's a more in-depth gig than you'd think.My family and I bought an abandoned 8-lane bowling center in the rural mid-west. In our small town there weren't many recreation options for families. You've heard of a food desert? This is an R&R desert.It had been abandoned for a good reason. The roof leaks, the electrical system was constantly surging, and my 70-year-old bowling equipment (still) doesn't work perfectly. The system that keeps your score is particularly interesting to me. It's the thing you watch during your game, but it fades into the background beyond that. Turns out these things are really cool, but absurdly expensive.Ours was installed in 2008 and cost six figures. It's calculating ball speed and trajectory, camera-based pin detection (object detection and trig, on ICs!), runs the fouling, the animations, the pinsetting machine and ball return. Very cool stuff for its age.From the business perspective, my facili
Ask HN: Does a local, Git-backed LLM "compiler" for personal notes make sense?Back in 2012, my very first startup was a note-taking and document storage app. Think Evernote meets Google Drive, but with a much faster UI. We managed to claw our way up to 1,800 active users. Then Google announced 25GB of free storage for Drive, and overnight, our business model evaporated. We had to shut down the company.I've been through the startup ringer four times since then. Despite the explosion of knowledge management tools over the last decade, I still manage almost my entire life and todo list in a single Markdown file. I’ve grown to severely distrust SaaS note-taking apps. I don't want my second brain held hostage by a monthly subscription, and I don't want it locked in a proprietary database.I'm thinking about building something new in this space, completely free and open-source, and I want a sanity check to see if anyone actually cares about this approach.The Idea: A lightweight, native desktop and Mac/Windows app that acts as an "LLM compiler" for your local files (ins
Ask HN: What leads to government websites being badly engineered?Government + public utilities (public transport, electricity, etc.).It seems this is an almost universal thing across all governments. Yes a few governments from UK and some Asian/European countries and even some in the US have recently been shipping good stuff.But by and large the APIs and the UI/UX tends to be much worse than what private enterprise tends to come up with.Consider for example booking a train in Spain. The official website of Renfe (the state owned railway company) is a mess. If you open the devtools and look at the network tab, you will get a cerebral aneurysm. This is true of other public transport systems around the world. Even the supposedly best one in Switzerland. For example if you go to sbb.ch and type Zurich, it will show you so many Zurichs that you'll have no idea what to select if you are new to the city (Renfe solves this with a "Madrid (ALL)" virtual entry). This leads to people using private wrappers around these like google maps/omio etc.I've traveled a
Atlanta Braves Announce Spencer Schwellenbach Update During Rangers SeriesThe Atlanta Braves are still clinging to a slight lead in the National League East. However, things have not been trending in the right direction, and a large factor in that has been the starting ...