I am actually using AI to learn about Windows OS performance. I have a couple of prompts scheduled to research, test, and implement performance gains in the OS, then post them here [0] and here [1]. I always admired Brendan Greg and Bryan Cantrill at Sun/Illumos/ for their work on OS performance but never had the time to sit down and catch up to their level due to my time in the military.
I got Codex to find and apply a bunch of settings I didn't know about and in the future I hope to use AI to make real changes in performance that venders probably don't have time or resources for.
that being said, what I've learned so far is essentially WhyIsItAlwaysHN's entire comment "The place to find them would be performance profiles, query plans, telemetry."
I just find it so frustrating that for a user interface that was basically solved in the 90s, Windows GUI still struggles to survive on hardware exceptionally more powerful than what we had 30 years ago
I am currently working on Clippy Vision, which is a fully local screen watcher that watches your screen continuously and builds context over time so you don't have to re-explain context like you do have to in case of Claude or ChatGPT.
Currently, its reactive, i.e a simple ReAct loop which calls function to retrieve data from memory and answers question when asked. Everything runs locally (Ollama + a local vision model, SQLite for storage), nothing leaves the machine.
Right now,I am planning to add a skill layer on top of the base watcher. So instead of it just being reactive, it acts proactively, for eg: spotting some mistakes in my code that I may have missed, or when you see X do Y types, and so on.
This month marks 1 year since I first started self-research and building my own coding agent VT Code [0]. I've been learning and experiment with a lot of things to first adapt myself and to learn about AI native/harness engineering. Thankfully, last month VT Code receives highest amount of contributions from the open-source community.
This month I will be focusing and enhance more long-running horizontal for VT Code such as plan mode, better strengthen tools policy structure and improve its harness as new and existing model grows exponentially. And If you are curious and want to fork it feel free, and welcome, let's build!
Agents have been huge for me. In addition to the Elm app, I used them to:
- build out ffmpeg/imagemagick image/video resizing
- unify all my servers behind one Traefik ingress
- build proper cross-site auth with a public/private key pair generated in the browser
- create Homebrew/macOS and Linux tarball distros of my service
- build an iCal calendar sync feature (if you browse the sites, you’ll see many of the events are synced from locally-oriented ICS feeds)
I also set it up to run my band’s new website at https://ato.band.
I just finished a loop generator program. I needed some sound effects (ambient noise, fire crackling, rain, waterfalls) but trying to hunt down perfectly seamless looping WAV files on the internet was fraying at my sanity.
So I created a tool where you can simply drag in a WAV/MP3 file, and it will automatically find suitable loop points, blend the transition, and export a seamless loop for use in games, soundscapes, ASMR, etc.
I'm currently updating my Common Thread game's dictionary with additional word combinations. It’s a game where you see three words and have to type in a word that pairs with all three (either in front or in back). For example:
I've got so many projects I'm simultaneously working on, but the one I really want to get over the line is "Todo driven development" - An edit to the gleam test-runner gleeunit that adds "todo" as a separate reporting category.
It feels really good when I've been trying it locally, but I've struggled to communicate the value to the maintainer.
Essentially, instead of having a bunch of failing tests that you fix to go green, you have a bunch of yellow "todo" tests, so that when you implement a part, you can see straight away whether you now get actual passing tests or failures.
Going from "12 failing tests, 16 passing" to "10 failing tests, 18 passing" doesn't tell you as much as "12 todo tests, 16 passing" to "1 failing test, 9 todo tests, 18 passing", or "10 todo tests, 18 passing". You can tell immediately whether a partial implementation was correct or not, rather than having to read the details of the failures to understand if they're from "todo" or not.
I'm working on Ekorbia: https://github.com/ekorbia/ekorbia-desktop. It's a lightweight, MIT licensed, Rust/Tauri desktop app for private local LLMs with a prompt library, chat overlay, ephemeral chat, model comparisons, and folder/RSS/URL watches.
I'm a movie guy and my friends and I play some daily movie frame guessing games. While fun, none of them are multiplayer or timed. For fun (and learning) Claude Code and I created a fast-paced, multiplayer movie guessing game over the past few weeks, HTML, Node, and PostgresSQL, leveraging the TMDB API. Feel free to check it out and drop me some feedback (link at the bottom of the game or here) on what you like, don't like, and if you see any issues. Enjoy!
I'm currently working on an open-source LLM inference engine [1] for browsers via webGPU. If anyone wants to try or has any suggestions, I'm happy to listen :)
I just finished up Denorm release 1 a couple of hours ago. It solves a class of biz logic problems I've faced since the 90s. Materializes derived columns in Postgres databases, within and across tables. Termination, no race conditions, non-subvertible, concurrency.
Our latest 5.0 release includes a save sync engines, so your game saves can move between devices (compatible app required). Start playing on your PC, save + sync, grab your AYN Thor and pick up where you left off.
I am building Bloomberry to help GTM teams find companies that started using over 1600 enterprise technology products (ie Hubspot or Workday) it’s an alternative to tools like Builtwith but focused much more on non-frontend tools
Please check it out and share your crtique - https://sharekeyx.com/
that being said, what I've learned so far is essentially WhyIsItAlwaysHN's entire comment "The place to find them would be performance profiles, query plans, telemetry."
I just find it so frustrating that for a user interface that was basically solved in the 90s, Windows GUI still struggles to survive on hardware exceptionally more powerful than what we had 30 years ago
0 - https://www.lacksan.com/updates/ 1 - https://github.com/Lacksan-Dev/HP-ZBook-Performance
Currently, its reactive, i.e a simple ReAct loop which calls function to retrieve data from memory and answers question when asked. Everything runs locally (Ollama + a local vision model, SQLite for storage), nothing leaves the machine.
Right now,I am planning to add a skill layer on top of the base watcher. So instead of it just being reactive, it acts proactively, for eg: spotting some mistakes in my code that I may have missed, or when you see X do Y types, and so on.
I am actively building this project. Version 1 is out already. source code: https://github.com/protocorn/clippy-vision
This month I will be focusing and enhance more long-running horizontal for VT Code such as plan mode, better strengthen tools policy structure and improve its harness as new and existing model grows exponentially. And If you are curious and want to fork it feel free, and welcome, let's build!
[0] https://github.com/vinhnx/VTCode
that is, to identify when papers have hallucinated references
can see an example of results at https://veruscite-data.com/share/3Pu3dwCfc-ixPgtHutVFXgMU7FY... for a preview. Also can sign up and get two free reviews.
Agents have been huge for me. In addition to the Elm app, I used them to:
- build out ffmpeg/imagemagick image/video resizing - unify all my servers behind one Traefik ingress - build proper cross-site auth with a public/private key pair generated in the browser - create Homebrew/macOS and Linux tarball distros of my service - build an iCal calendar sync feature (if you browse the sites, you’ll see many of the events are synced from locally-oriented ICS feeds)
I also set it up to run my band’s new website at https://ato.band.
So I created a tool where you can simply drag in a WAV/MP3 file, and it will automatically find suitable loop points, blend the transition, and export a seamless loop for use in games, soundscapes, ASMR, etc.
Free and open source.
https://github.com/scpedicini/loopweave
I'm currently updating my Common Thread game's dictionary with additional word combinations. It’s a game where you see three words and have to type in a word that pairs with all three (either in front or in back). For example:
• butter
• fishing
• ball
https://common-thread.specr.net
It feels really good when I've been trying it locally, but I've struggled to communicate the value to the maintainer.
Essentially, instead of having a bunch of failing tests that you fix to go green, you have a bunch of yellow "todo" tests, so that when you implement a part, you can see straight away whether you now get actual passing tests or failures.
Going from "12 failing tests, 16 passing" to "10 failing tests, 18 passing" doesn't tell you as much as "12 todo tests, 16 passing" to "1 failing test, 9 todo tests, 18 passing", or "10 todo tests, 18 passing". You can tell immediately whether a partial implementation was correct or not, rather than having to read the details of the failures to understand if they're from "todo" or not.
I"m also still working on my crossword puzzle pipeline, https://github.com/ekorbia/xword-pipeline, and my daily word game site, https://wordfuzz.com/, which has a daily mini crossword and a few other daily word games.
And finally I'm working on a browser based Civ5/Unciv clone which is in its early stages.
https://playflashframe.com/
1. https://github.com/stfurkan/bitgpu
Demo: https://stfurkan.github.io/bitgpu/examples/chat.html
2. https://aidekin.com --> this is one of my project that's currently using bitgpu engine
https://github.com/Fluid-CAD/FluidCAD
https://gitlab.com/kendowns/denorm
Apache license.
- Camelana. A proportional coding font with custom ligatures for `return` and `throw` keywords
- Tabular Eye. IDE plugin for tabularizing mono and non-monospace fonts without adding whitespace
https://github.com/ericfortis/camelana
https://github.com/ericfortis/tabular-eye
Our latest 5.0 release includes a save sync engines, so your game saves can move between devices (compatible app required). Start playing on your PC, save + sync, grab your AYN Thor and pick up where you left off.
https://bloomberry.com