Is anyone working on or knows a library for evaluating LLMs for application features and/or application features that use LLMs? I am wondering what people use or if anyone has their own solution.
I'm working on Argon Chess, a deterministic chess variant with some degree of cheat resistance (hard to describe to chess engines like Fairy Stockfish) and tons of variety. A week ago, I added a way to play friends online a week ago (a Discord Activity) and a simple Play a Dumb AI feature on its website. You can also print the cards for free for offline play. https://argonchess.com/
Discovered in-door bouldering / rock climbing and now go 3x a week, am absolutely loving it! Because of that, I haven't really worked on any side projects in a while. Perhaps I don't need to? My job advances me plenty in my field, but it is a bit of a bitter-sweet feeling in a sense, like maybe I should try to squeeze more out of my free time somehow.
I climb a lot! (Actually currently sitting on Big Sur ledge on el cap posting this).
It cuts into my free time programming for sure, but imo super worth it! Enjoy it, it’s a wonderful hobby.
Why? I love the old arcade and game boy games, and I want to recreate them to my liking. I also love mechanical systems and space rovers, and I want ro build worlds to explore and simulate these things
Frontend framework in JavaScript that requires no build step, relies on DOM and SSR and can be used to build both SPA and hybrid apps without VDOM, js templates, hydration or putting HTML (or worse, css) inside JS code. It'll also have a very sophisticated declarative state manager which makes managing state and ui transitions a breeze. It's basically anti-React.
Been nerd sniped recently so am working on a Rust version of markdownlint-cli2. I'm tired of having a node dependency in my projects and this seems like a constrained enough problem space that I'll actually get around to doing it.
Working on https://ziva.sh/, an AI agent for game development. It uses MCP to integrate with Godot, a leading open source game engine.
It's coming together really nicely, targeting a beta release later this month. If anyone is interested in game development and wants to be a beta tester, lmk :)
I am working on methods to automate my VC firm. We have a small team and many different tasks to do. I’ve had success with using LLMs to help us automate various projects. But I appreciate any open source tools, techniques, readings, etc. if anyone knows any!
Unlike traditional accounting platforms we expose the ledger model directly which enables our customers to model complex transactions even when we do not have direct support for it.
Been working on this for a month, and it uses Elixir, Phoenix and InertiaJS with React.
I rewrote Playwright to run completely in a Chrome Extension without CDP or chrome.devtools for no practical reason at all. I started to do it like Forest Gump started running. It can't get past bot protection so pretty worthless from a browser automation point of view. [0]
What I don't understand is why the need to rewrite Playwright instead of just patching it. Playwright (or Puppeteer) has addressed every edge case that has come -- especially race conditions which are a monster to deal with -- up over the years and by the time you do the same you will have Playwright.
Why is rewriting or rebuilding Playwright from the ground up needed?
I recently added FSRS (besides also having Anki integration). Now I'm working on replacing the need for reviewing flashcards by having reading activity automatically mark flashcards (current and future) as reviewed, so that you can get many of your reviews in just by reading native materials that interest you instead of sacrificing most of your study time to contextless flashcard grind.
I'm also working on a manga mode using a new manga OCR tech I have licensed out of academia that is ahead of state of the art alternatives.
Building https://www.hessra.net/, an authorization system based on the Biscuit token format (decentralized, signed, and attenuable). The goal is to push beyond JWTs and Zanzibar-style policy engines by giving every machine-to-machine request its own embedded, verifiable authorization logic in a small capability token. These tokens can be delegated, restricted, and verified locally with no extra network calls required after getting the token.
Early use case is replacing API keys with identity tokens that expire, delegate, and prove possession and then can be used for easy step up to fine-grained authorization. There's some pretty interesting authorization stuff you can do, like having multiple parties sign off before a token is valid or requiring a series of micro-services sign a token for it to be valid.
A kernel extension-less sshfs for macOS. I tried using FSKit and got halfway before I felt too constrained by the extension security model (must be app sandboxed, must be approved by the user in system settings). Now it’s just a standalone command line binary that doesn’t require any special permissions since it proxies NFS to SFTP. Everything “just works” and performance is reasonable
I’m working on Reflect [0], it’s a privacy-focused app for self-tracking and self-discovery. You can track metrics, run self-experiments, set goals, view correlations, visualize your data, etc.
I posted in this monthly thread first time in May when I launched a daily logic puzzle, Clues by Sam. Since then it's grown significantly, and I couldn't be happier!
The game has a farily simple frontend, but there is a fairly complex constraint solving algorithm as part of the puzzle making process. What makes the puzzle quite unique is that you can't "guess". You can only make guesses that are provable by logic. The algorithm ensuring this has worked flawlessly for months now (though I've manually inserted some silly mistakes once or twice).
Today's puzzle is one of the hardest to date. The difficulty resets on Mondays, and then gets harder again towards Sunday.
It looks inside each file to see what it’s about, then moves it to the right folder with a single click.
Everything happens on your Mac, so nothing leaves your computer. No clouds, no servers.
It already works with images, Office (Word, Excel, PowerPoint) PDFs, ePubs, text, Markdown, and many other file types (30+) in English. Next I’m adding multi-language support.
If you have messy folders anywhere on your Mac, Floxtop can help.
Working on https://gametje.com (a Jackbox games competitor). Been working on the Android TV app lately. Will probably start creating a new game next week with acronyms similar to the old game Acrophobia from the late 90s/early 2000s.
I've been building a little toy computer and assembly language that's interpreted in python. Pretty close to the first release (and introductory blog post) and a lot of fun to build (and learn a bit more about real assembly as I go).
It's an explorable database of films, TV shows, books and board games based around their historical setting: where and when the thing is set. It's been incredibly complex and interesting getting the (messy) data, making sense of it and trying to design a UI to explore it.
1,005 comments
[ 3.1 ms ] story [ 201 ms ] threadAlso planning on adding more tools to help development teams.
I am almost done with flappy bird (2nd challenge)
Why? I love the old arcade and game boy games, and I want to recreate them to my liking. I also love mechanical systems and space rovers, and I want ro build worlds to explore and simulate these things
Thinking on good export formats (except of taking screenshots and Pull Requests, obviosuly). LaTeX and Typst? A remark plugin?
A law professionals helper - aggregates judicial case info into a single place, gives visibility and notifications - asistentul.ro
A scheduling platform for self-employed professionals that offer services (think hair-cutting, nails, psychlogists). (Not yet live)
Aaand something in compliance that I want to keep a bit stealthy right now.
It's coming together really nicely, targeting a beta release later this month. If anyone is interested in game development and wants to be a beta tester, lmk :)
Unlike traditional accounting platforms we expose the ledger model directly which enables our customers to model complex transactions even when we do not have direct support for it.
Been working on this for a month, and it uses Elixir, Phoenix and InertiaJS with React.
It's called Stagehand (https://github.com/browserbase/stagehand) and we just released v3, which is a total rewrite.
I rewrote Playwright to run completely in a Chrome Extension without CDP or chrome.devtools for no practical reason at all. I started to do it like Forest Gump started running. It can't get past bot protection so pretty worthless from a browser automation point of view. [0]
What I don't understand is why the need to rewrite Playwright instead of just patching it. Playwright (or Puppeteer) has addressed every edge case that has come -- especially race conditions which are a monster to deal with -- up over the years and by the time you do the same you will have Playwright.
Why is rewriting or rebuilding Playwright from the ground up needed?
[0] https://github.com/adam-s/cordyceps/tree/main/pages/side-pan...
https://reader.manabi.io
I recently added FSRS (besides also having Anki integration). Now I'm working on replacing the need for reviewing flashcards by having reading activity automatically mark flashcards (current and future) as reviewed, so that you can get many of your reviews in just by reading native materials that interest you instead of sacrificing most of your study time to contextless flashcard grind.
I'm also working on a manga mode using a new manga OCR tech I have licensed out of academia that is ahead of state of the art alternatives.
This project now sustains my full-time focus.
Early use case is replacing API keys with identity tokens that expire, delegate, and prove possession and then can be used for easy step up to fine-grained authorization. There's some pretty interesting authorization stuff you can do, like having multiple parties sign off before a token is valid or requiring a series of micro-services sign a token for it to be valid.
[0] https://apps.apple.com/us/app/reflect-track-anything/id64638...
The game has a farily simple frontend, but there is a fairly complex constraint solving algorithm as part of the puzzle making process. What makes the puzzle quite unique is that you can't "guess". You can only make guesses that are provable by logic. The algorithm ensuring this has worked flawlessly for months now (though I've manually inserted some silly mistakes once or twice).
Today's puzzle is one of the hardest to date. The difficulty resets on Mondays, and then gets harder again towards Sunday.
https://cluesbysam.com
It looks inside each file to see what it’s about, then moves it to the right folder with a single click. Everything happens on your Mac, so nothing leaves your computer. No clouds, no servers.
It already works with images, Office (Word, Excel, PowerPoint) PDFs, ePubs, text, Markdown, and many other file types (30+) in English. Next I’m adding multi-language support.
If you have messy folders anywhere on your Mac, Floxtop can help.
Try it for free - requires macOS 14+ with Apple Silicon: https://github.com/taranntell/fallinorg/releases/download/1....
Just for clarity, it looks like image content itself isn't addressed, but rather just any text that might be in an image, correct?
Also: "Your sensitive data never leaves your Mac." Does anything leave the mac? Any metrics? I don't want this to have network capabilities at all.
https://github.com/daturkel/dt31
It's an explorable database of films, TV shows, books and board games based around their historical setting: where and when the thing is set. It's been incredibly complex and interesting getting the (messy) data, making sense of it and trying to design a UI to explore it.
https://stevebennett.me/2025/08/26/whenever-exploring-times-...
I will note that there are some Norfolk England things showing in Norfolk Virginia.
https://github.com/willtobyte/carimbo