In the real world finally moved everything to USB-C. Gave all my old cables away. I have two chargers in my home and a handful of C to C cables. Everything connects to everything now.

Home is now downgraded to a dumb home. Lights work on physical toggles. No hubs or sensors anywhere. Heat and AC is with a dumb panel on the wall.

It feels freeing.

It feels like somewhere in the last decade we've all lost control over our email inboxes. While it would certainly be possible to filter and sort it, I've been wondering if it makes sense to just start with a system that is designed to intake a bunch of streams of information. Then it could be pointed at the raw information e.g event calendars and news-letters as well as streams like Facebook groups/Instagram where I don't want to actually go to those apps.

Speaking at a meta-level, this seems like what we should really be using LLMs for right now: use-cases where user controls what is done on their behalf.

Completely free, no ads, no in-app purchases and no accounts / network required offline voice transcription.

I have also built the macOS/Windows/Linux versions which I'll also make free to download and available on my site soon (https://blazingbanana.com/).

iOS version is built and works (extremely well), just waiting for the Apple Developer signup process to complete.

Big shout out to https://github.com/mybigday/whisper.rn and https://huggingface.co/ggerganov/whisper.cpp/tree/main for making this even possible.

Any suggestions are welcome.

See https://mediareduce.com

Feedback welcome

It started out as something marginally more useful than vendoring your dependencies as submodules + baking in the knowledge of how to build a bunch of common projects.

I realized, though, that there was somehow a huge gap in the insane world of C build tools. There's nothing that:

- Lets you pin really precisely and builds everything from source (i.e. no binary repository)

- Does not depend on either a scripting language or a completely insane DSL (Conan uses Python, CMake is an eldritch horror, ditto Make, lots of other tools of course but none of them quite hit the mark)

- Has a good balance of "builds are data" and "builds are code".

Anyway, it's going great. There are, of course, a ton of problems to solve. Chief among them is the obvious caveat that C is not a monoculture like Rust. There will be zero upstream libraries that use this tool natively. But I don't think it matters. I think I can build something which is as much better to the existing tools as, say, UV was to existing Python tools, even with that disadvantage.

An PWA primarily for my wife and my daughter. They can order their hot chocolate and their coffee as if they were going to grab something at a fancy café downtown, but instead it's at home and I'm the barista. It is quite nice to have for when my wife comes back from work and want something specific, or when we are waiting for the visit of a few friend, they can order exactly the available beverages and everything is ready when they're here.

It was also a good playground for me to implement Web Push notifications (to never miss new orders).

It's a basic Nuxt 3 app with Appwrite as the backend with rough edges, but much enough for our household use !

If you want to spam my phone with notifications, please visit my café : https://mytinycafe.com/alix

Remixify automates the search while leaving the selection to you. You paste a Spotify playlist URL, and it helps you or provides you a good starting point for digging. It groups the results by the original track so you can quickly preview and save the versions you want to a new playlist.

We don't try to recommend new music or use AI to guess your taste. It just finds the usable versions of the music you already selected.

[1]https://github.com/kwakubiney/remixify

[2]https://remixify.xyz/

It fetches new papers, scores them against a “research profile,” then produces concise summaries plus a short “why this matters” style rationale, and outputs an email/newsletter-like HTML digest. There’s also a small API for generating a digest, checking status, and previewing the render.

I built it because keyword alerts and generic newsletters were either too noisy or missed the stuff that was actually relevant to what I’m working on right now.

https://iotdata.systems/jsonlviewerpro/

Next step is to integrate a visual data pipeline by using ImNodes. I‘m slowly making progress in my experiments, but C++ has a steep learning curve, especially when targeting MacOS and Windows at the same time.

I’m still early and adding ideas as I go, but it’s already helped me questions I had.

Examples: - Coin flip simulation: https://www.blockviz.xyz/simulation/coin-flip - Sell & buy-back simulation: https://www.blockviz.xyz/simulation/sell-buy-back

Curious if others here run into similar “this felt right, but did it actually help?” questions.

Here is a work in progress build:

https://muffinman-io.itch.io/space-deck-x

It is a combination of a shoot-em-up and deck building. You fly and shoot until you get to the boss, when you get your deck out to fight them.

That genre combination is definitely too ambitious, but I think it is fun to play and I’m enjoying making it.

I have a bunch of ideas how to combine the two parts better. But over the years, I’ve learned to control scope creep and actually ship pet projects.

Right now I’m in a middle of changing how enemy waves are spawned. After that I want to make a short tutorial and add two more bosses as well as more enemies.

If you end up playing it, please share your feedback I’ll be glad to hear it.

The game is made using Kaplay, a game dec library which brings me joy to use. I can best describe it as my friend described Pico-8: “easy things are easy”. But compared to Pico-8, Kaplay doesn’t have virtual console limitations and comes with a big library of components. Try it out, the community is small, but the library itself is really fun and easy to use.

EDIT: For context, this is about two weeks of work, in the evenings when my kid is asleep.

Right now I have a build that loads in the browser, but I really want to have "multithreading" which means workers in the web. One can use asyncify with emscripten to translate blocking C++ to WASM, but that transition is not perfect, right now I'm debugging a bug where there's a race condition that halts all execution and the main thread runs in an infinite loop waiting for the workers to stand up. I guess I'll have a few of those ahead.

The main goal is to 1. just have fun 2. use yjs as a collab backend so multiple people can edit the same PCB. This will probably work with pcbnew, KiCad's layout editor, since it has a plugin system and AFAIK I can do the sync layer there. For the rest ( schematic, component editor etc. ) I'll have to figure out something.. KiCad does not sync automatically if you modify a file, I'll have to do some lifting there.

Anyway, it's a lot of fun, I really want this thing to exist, I'm hoping that I won't run into a "wellll, this is just not going to work" kind of issue in the end.

A big part of this was inspired by the last startup I worked at. In an effort to not deal with complexities of Kubernetes, we ended up on Heroku and was charged exorbitant amounts of money. One year spending close to 400k on Heroku alone, for what should’ve been 10-15k in cloud costs.

I think a big part of this is just making Kubernetes more friendly and easier to use for a small / midsized team of developers.

The goal is to make it easy enough for even a single developer to feel comfortable with, while also being powerful enough to be able to support a small team

I made a daily word puzzle called Tiled Words.

https://tiledwords.com

Currently about 2,000 people play every day and I’ve released 59 puzzles!

One feature I’m excited about is crowdsourcing puzzles. Today’s puzzle is a “community puzzle” made entirely from clues that players submitted! I plan to do this every week or two.

I wrote about launching and the first month of puzzles if you want to learn more!

https://paulmakeswebsites.com/writing/a-month-of-tiled-words...

Since I was researching DNS and global mobility, and wanted to share links with others, figured I'd just spin up a link site (though I'm still the only user).

One unique difference is I have a field for English Title, since I consume a lot of Korean & Japanese articles and want to share these, but don't want to have people translate the titles before they understand why they should read them.

https://news.reorient.guide/

I see it as a "poor man's continual learning".

I’m building it on Cloudflare Workers with advanced tracking, modern templates, and advanced webhook integration. Developers can also configure and schedule advanced workflows for their specific needs

The users can review their usage and performance using an intuitive dashboard.

Email is a crowded space and this is my first attempt at doing something indie at this scale. Wish me luck!

- a videogame. I've got a pretty killer idea in an open niche, but the indie market is so massively oversaturated that it feels impossible to get eyeballs.

- a next-generation post-RSS newsreader. But news is so depressing these days. I think most of the world wants to ostrich and I don't blame them.

- a reboot of Svpply, my own shuttered startup. I'd love to just make (another) thing that's about excellent clothes and shoes and artisanal pocketknives, but the way the economy is going, this feels grotesque. I was lucky to make it the first time when luxury goods were attainable _and_ normal people could pay for necessities; that window has closed.

USCIS forms still use XFA PDFs, which don’t let you edit in most browsers. Even with Adobe, fields break, and getting the signature is hard.

So I converted the PDF form into modern, browser-friendly web forms - and kept every field 1:1 with the original. You fill the form, submit it, and get the official USCIS PDF filled.

https://fillvisa.com/demo/

What Fillvisa does:

- Fill USCIS forms directly in your browser - no Adobe needed

- 100% free

- No login/account required

- Autosave as you type

- Local-only storage (your data never leaves the browser)

- Clean, mobile-friendly UI

- Generates the official USCIS PDF, ready to submit

- Built-in signature pad

I just wanted a fast, modern, free way to complete the actual USCIS form itself without the PDF headaches. This is a beta version

It started as something I wanted to build for myself. I have a Bosch dishwasher that lacks any glanceable indication of how far along it is. Bosch provides an app, but checking the progress takes too long to be useful.

I figured live activities was a good fit, and then realized that I am not alone in wanting something like this. So, I am trying to make it into something usable for all the home automation tinkerers.

https://getaivi.app

Since hacker news last saw it, it’s been translated into English, German, Spanish and Chinese. If, say, a Chinese speaker wanted to learn more English words, then they could go to https://threeemojis.com/zh-CN/play/hex/en-US/today and play the game with English words with Chinese definitions and interface. This is the first cross language daily word game of its kind (as far as I know), so it’s been a lot of fun watching who plays which languages from where.

The next challenge that I’m thinking about is growing the game. The write ups and mentions on blogs add up, the social sharing helps, but I’d really like to break into the short form video realm.

If you read interviews from other word game creators, every successful game has some variation of got popular riding the wordle wave, or one random guy made a random TikTok one time that went super viral, and otherwise every other growth method they have tried since then hasn’t worked that well and they are coasting along.

So, sans another wordle wave, I am working on growing a TikTok following and then working on converting that following into players, a bit of a two step there, but that’s how the game is played these days. https://www.tiktok.com/@three_emojis_hq for the curious. Still experimenting and finding video styles and formats that travel well there. Pingo AI and other language apps have shown how strong TikTok can be for growth, so I think there’s something there. That’s all for this month!

https://github.com/btrettel/blastersim

The core simulator part works, but I don't yet have a user interface or documentation. Probably just going to be text input files to start, maybe a GUI later. Recently, I'm mostly working on testing.

The simulator is object-oriented and basically allows one to build up a blaster from separate control volumes and connections between control volumes. This is useful as it allows the same core simulator framework to handle different blaster configurations and even variants of them. For example, someone asked me to make the spring piston able to pull a vacuum on its back side due to not having sufficient flow. That's easy here as I just need to add another control volume and the appropriate connection onto the basic springer configuration.

Store your graphs in Parquet files on object storage or DuckDB files and query them using strongly typed Cypher. Advanced factorized join algorithms (details in a VLDB 2023 paper when it was called Kuzu).

Looking to serve externalized knowledge with small language models using this infra. Watch Andrej Karpathy's Cognitive Core podcasts more details.

https://github.com/MarkusSintonen/pyreqwest

- Updating my personal SSG to support Obsidian fully, which should simplify the publishing process a bit more. https://0xff.nu/hajime/

- Trying to find a new job, which is proving to be more difficult than it should be if you have certain standards about work/life balance.

- Writing an informative article about automating with/for ADHD which explains the motivation and solutions that I came up with for perhaps the weirdest, yet most annoying issues I face or forget about on a daily basis.

All notes are simple markdown file stored locally.

I’ve been using it to benefit my research and make the knowledge to stick better on my head for several years. My base is more than 400 markdown notes now, and I sync them to a private GitHub repository.

https://github.com/odosui/mt

In other words, something safer & more concise than maintaining multiple HashMap's, but a lot less involved & simpler than an in-memory SQLite.

It's better explained by the example here: https://github.com/utdemir/composable-indexes/blob/3baa36762....

https://github.com/utdemir/composable-indexes

https://mydragonskin.com/products/stingray-treatment-kit

This will help people set clear expectations for their apartment search.

[0]: https://allaboutberlin.com/tools/rent-map

Two main differences between this and other Anki-like apps: 1) The words you learn are from YT videos, websites and ebooks you import in the app. 2) The flashcards are optimized specifically for learning vocabulary - cards automatically get audio, images, multiple sentence examples, words definitions etc. It can also create fully monolingual flashcards with just definitions or the words in dialogs.

My biggest flex is that I have users who have done more swipes than me (over 100,000).

1) https://github.com/radusuciu/snakemake-executor-plugin-aws-b... (my fork). Just add the features to the batch job building code 2) https://github.com/radusuciu/snakemake-executor-plugin-aws-b.... This is more experimental and not yet fully working. I wanted to try a few things. a) can we rely on existing job definitions (managed through IaC instead). b) can we implement a fire-and-forget model where the main snakemake process runs on Batch as well? c) Can we slim down the snakemake container by stripping off unnecessary features.

The first is a customizable digital math workbook. Currently the demo covers fourth grade math. There is a practice mode where you can select the skills you to want practice. There is also a customizable dashboard where you can setup your own widgets to practice math skills in different ways. I am working on some pre-made dashboards to help users get started. The next plan is to cover fifth grade math skills. My plan is to cover first grade math up to Calculus and High School Physics. I envision it as a companion tool for Khan Academy/Math Class/Math Books. Check out the demo. No signup required. Progress is only stored locally.

https://demo.numerikos.com/

The second thing I am working on is an application to practice Cangjie. It's a Chinese input method that has been around for a long time. It is based on a visual decomposition of characters. Each character is represented by one to five codes and the majority are unique. My application teaches Cangjie like keyboarding (QWERTY) is taught to young students. You learn the location of the keys, then some basic words, then start typing sentences. I also have a free demo for it as well.

https://demo.cangjieworkbook.com

Feedback on either project would be appreciated.

This looks like keyboard driven commands, secrets store (to be done) and scripts that you can write and store without spinning up a new server (easier chat ops)

Still in early alpha so after a few more polish it'll be ready, but you can try it right now!

Currently trying to better contextualize the visible subregion of the matrix in relation to the full dataset (beyond what the current minimap does).

https://cybernetic.dev/matrix

- scenes composed of SVG shapes, text, etc.

- web-worker rendering everything on the offscreen canvas;

- elements positioned via yoga-layout;

- optional JSX layer to define layouts, no support for React components inside the layout (yet);

- using Skia now, maybe Rive Renderer / Vello later? — I'd love to migrate to WebGPU eventually,

- first-class view transitions: no white screen, no jumps after the initial load, no things appearing/disappearing without a proper transition);

- fontkit to calculate everything re fonts and shape text — no more DOM-provided measurements;

- integration with Remotion to render videos.

Short-term goal is to reach MVP for slides/dataviz tool, and I'm getting close.

Trying to stay at maximum FPS while sacrificing loading time and, sometimes, the battery life.

It uses LLMs to generate python code to scrap a webpage to fit any Pydantic model provided:

  from hikugen import HikuExtractor
  from pydantic import BaseModel
  from typing import List
  
  class Article(BaseModel):
      title: str
      author: str
      published_date: str
      content: str
  
  class ArticlePage(BaseModel):
      articles: List[Article]
  
  extractor = HikuExtractor(api_key="your-openrouter-api-key")
  
  result = extractor.extract(
      url="https://example.com/articles",
      schema=ArticlePage
  )
  
  for a in result.articles:
      print(a.title, a.author)

* The immediate-mode "every tick I ask you for a VDOM based on the user-defined state" TUI framework has all the fundamental features, I think; writing docs and expanding the library of components it ships with. https://github.com/Smaug123/WoofWare.Zoomies

* Decided I needed a nice text display widget, so got side-tracked into implementing the Knuth-Plass paragraph layout algorithm; it currently functions but is buggy. https://github.com/Smaug123/WoofWare.KnuthPlass

* Finally starting to put proper effort into the LLM integrations into my workflows, writing skills, defining the Gospel According To Me to try and poke the LLMs into the right basin - with limited success so far. https://github.com/Smaug123/gospel

No progress on the deterministic .NET runtime.

(Same comment from last month: https://news.ycombinator.com/item?id=45869787)

Seriously, I'm very proud of myself for the little I've accomplished so far. I don't have friends in tech so I don't get to talk about it or bounce ideas off people.

Thanks for letting me get that out!

If you have played military sim (Milsim) games like Project Reality, Squad or Arma you might appreciate it.

Its quite cool how the game devs have made a lot of tooling to use; they use Typescript to hook into in-game events and functions.

There is a whole community making lots of content too:- https://bfportal.gg/

Currently I am working on an insurgency game mode; where one team has to defend some caches and use guerilla tactics, whilst the other team has a smaller size but the advantage of firepower and vehicles.

Hopefully have it released by Christmas time.

Then I wrote a Python program that connects whatever controller my brothers want to use (as long as it's supported by SDL2.0) and forwards that data from their computer, through Parsec, through a USB-UART adapter, to the Pico, then to the Switch. I then have a low latency capture card (Magewell Pro Dual HDMI I got off of ebay for $100) forwarding the video and audio from the Switch to my PC which I share to my brothers via Parsec. The audio was a bit tricky to get right, and ended up having to use a Virtual audio cable and Voicemeeter potato (a software audio mixer) so that both myself and my brothers could hear the audio.

It works surprisingly well and the latency is pretty low. I even got rumble working! (but not motion controls. If anyone wants to attempt it, I will accept PRs). I haven't done any formal benchmarking for performance, but my brothers and I were able to play Smash Ultimate without too much bother about latency.

You could also use the accessory Python library I made to automate switch controller presses (look in the examples directory). Might be useful for TAS speedruns?

The project is here for anyone interested. It's a bit rough and needs some cleanup and maybe a video tutorial on remote setup. But here is the WIP:

https://github.com/jyapayne/switch-pico

1. probe.bike - tell stories with your bike rides. It allows you to aggregate your cycling trip into one datapoint. Will likely break this out to skiing over the break and rebrand slightly. Adding yearly cards as we speak!

2. flopper.io - I'm seeing traffic rise and rise for this and it's been a great way to translate my every-increasing understanding of AI Infrastructure architecture to a new project. It acts as a benchmark website for GPUs and systems (e.g. Nvidia NVL72.

3. llmstxt.studio - still feel like llms.txt as an idea make sense - so hedged that and but let's see. Got my first customer this month. B2B and need more features/marketing.

4. rides.bike - the oldest - a catalogue or well researched cycling destinations and information about destinations. Will be adding more very soon!

Place discovery companion that de-noises your environment. Repeatable, one-stop-shop for information, personalized. Quick to decision. Updates live (best on mobile).

--

We are passionate travelers with 30k km under our wheels and we want consistent information across places we find ourselves at. Now are trying to figure out how to help others.

The main window uses Apple’s local LLM to summarize your conversation in realtime, with some swoopty UI like QUEUED state on Claude Code.

I’ve just added macOS Sequoia support and a really cool CLI with Claude Code skill allowing seamless integration of information from your conversational history into aI’s responses to questions about your development history.

The CLI interface contract was designed to mutual agreement between Claude code and codex with the goal of satisfying their preferences for RAG.

This new query feature and pre-Tahoe support should be out this week, but you can download the app now on the App Store or as a DMG.

I’m very excited about this App and I would love to get any feedback from people here on HN!

https://contextify.sh

My Show HN: from this past week has a short demo video and a bit more info:

https://news.ycombinator.com/item?id=46209081

Currently spending time establishing relationships with historical societies, as I really need them to contribute points of interest, and stories. Many of these societies are run on a voluntary basis by 70+ year olds, so it's a long process. Getting some good responses eventually though, so it might actually go somewhere, just a lot slower than I want.

Also still doing https://wheretodrink.beer, but haven't added anything of note since playing on this other project.

And react2shell was a blast

https://github.com/0xekez/tinyLIRPA

tinygrad’s small set of operations and laziness made it easy to implement. Tho my overall sense is that neural network verification is currently more of a research interest than something practical.

The reason for it was because after testing multiple Git history explorers, I still think nothing beats the gitk. Sublime Merge is probably the only alternative that I would seriously consider but I don't really like the UI and the fact that it is proprietary (I am not against proprietary software but I prefer an opensource solution when available). Other alternatives have some bugs or the interface few too slow. gitk itself is mostly fine, but sadly it tries to load the whole repository in memory and this is causing issues every time I try to navigate through nixpkgs (I can see the memory consumption going through the roof while the UI slow down to a crawl).

gitk-go loads a batch of commits (1000 by default) and once you get at the end of the list it loads more. I also add a few features that I miss from gitk, for example if you do any change in the repository (change branches, add files to stash, etc) it will automatically reflect in the UI.

Again, the code is mostly vibecoded since this is the first time I decided to try this from scratch. The code works well for my use cases and it is enough to replace gitk for me, but I can't guarantee there is no bugs and the amount of tests are small. But still, it was fun to see something that I wanted to create for a while (I had this idea for a long time since the issues with gitk that I was having) finally taking form. Probably the program is not useful for anyone but me, but if anything this is a feature, not a bug.

None of the frontier LLMs (Gemini, ChatGPT, Claude) produce usable designs when just prompted with some photos of the pump and a written description of the mount. I'm now building a simulator in Mujoco that the LLMs can use to test and iterate on their designs to see if they can do better in this setting.

I'm hoping to make an interesting blog post of it and maybe end up with a usable wall mount design.

Next years (and probably a couple years after) is an electro-mechanical smart watch. Sourced some Ronda GB22 gearbox motors and tritium tubes and planning on using a pcb for the face. What could go wrong.

Working on a new puzzle for it as well as the mobile app, which is coming for iOS and Android around the holidays.

https://donethat.ai

With lot's of built-in data privacy safeguards https://donethat.ai/data

Also made an overview of similar tools out there https://donethat.ai/compare

Recently broke on Linux with a Wayland security update, working on a fix! Using Electron for cross-platform.

You can think of it as a data source, or a knowledgeable companion that can provide comprehensive book information for online booksellers, libraries, book-related startups, bookworms, and more.

I got a pre-alpha build running for those that want to test it out and the code is out on SourceHut[1].

Been really tough to find time to work on it because I have a baby that only sleeps in my lap, but I’m making progress very slowly.

I recently hired someone to rewrite the entire database layer, as that was written with the help of an LLM for the prototype, which should improve things too.

Feedback is very welcome :)

[1]: https://sr.ht/~pagina394/librario/

I’m only a couple days in, and I’ve already learned so much about networks, containers, codecs, ffmpeg, and so on.

https://www.inclusivecolors.com/

There's 100s of color palette generation tools, where most only let you customize a single color then try to autogenerate tints/shades without much thought about accessibility or tints/shades customization. The main features of this tool are:

- Emphasis on accessibility. A live UI mockup using your palette warns you if your tints/shades are lacking contrast when used in practice for headings, paragraphs, borders, and buttons, and teaches you the WCAG rules. Fixing contrast issues and exploring accessible color options is also made much easier using an HSLuv color picker, where only the lightness slider alters the contrast checks, and not the hue/saturation sliders (most tools use HSL, where hue/saturation changes counterintuitively alter contrast checks which makes accessibility really tough!).

- You can tweak the hue/saturation/lightness of every tint/shade. This is useful because autogenerated colors are never quite right, and customization is really important for branding work when you have to include specific tints/shades. The curve-based hue/saturation/lightness editing UI also makes this a really quick process.

- Instead of just a handful of colors, this tool lets you create a full palette. For example, if your primary color is blue, you always end up needing other colors like green for success, red for danger, and gray for text, then 11 tints/shades for all of these, so you want a tool that lets you tweak, check, compare and manage them all at once.

It's mostly a demo on mobile so check it on desktop. I'm still working on making it easier to use as it probably requires some design background to understand, but really open to feedback!

It helps to comprehend research papers (and not only papers - any document on any language) faster.

The tool is free to use, because we have credits from GCP. I guess at some point we'll need to introduce some level of subscription fee to keep it alive and useful, as it uses LLMs and vector search quite a bit.

Feedback is welcome!

Currently building out support for multi-agent evals, better tracing, voice, and static code analysis for AI security use cases. So many fun sub-problems in this space - LLM testing is deceptively hard.

If you end up checking it out and pick up an issue, I'll happily send swag. We're also hiring if you want to work on this stuff full-time.

https://github.com/promptfoo/promptfoo

You get to choose the genres you're interested in, and it creates playlists from the music in your library. They get updated every day - think a better version of the Daily Mixes. You can add some advanced filters as well, if you really want to customise what music you'll get.

https://riffradar.org/

Along the way I found most of these use salvaged BlackBerry keyboards which are only going to become harder to find, so also on a bit of a side quest to build a thumb-sized keyboard from scratch. Got me into laying out and prototyping my first PCBs and learning about how these things are made - lots of fun so far!

Something cool I learned from tearing apart a BB keyboard: the satisfying “click” is just a tiny metal dome that pops and completes the circuit when pressed. Not news to anyone familiar with electronics manufacturing, but it was a cool thing to “discover.”

- Arduino dev and circuitry

- 3D printing

- PCB design

- Woodworking

Its all a lot of fun and IMO a lot more approachable than it has been thanks to the assist from LLMs.

- Just finished forking an nvim keycast script for TUI demos: https://github.com/wong-justin/showkeys-noplug

- Started making a Roku app (https://wonger.dev/nuggets#n299)

- Drafting a year-in-review post for my website

- Drafting a book review for "Programmers at Work"

https://github.com/TechSquidTV/Tuvix-RSS

Eidetica - a decentralized database built in Rust, intended for local-first apps. It's still unstable but I'm progressing relatively rapidly. In the past ~month I have:

- Flown to SF to attend a conference in this niche: https://syncconf.dev/

- Added password based, transparent, end-to-end encryption

- Improved my custom CRDTs

- Added an index to store configs and metadata

- Built support for using sqlite + postgres for Eideticas backend (not pushed yet)

Once I finish the backend work I'll hopefully take a bit of a break though. I'm supposed to be retired.

Will be interesting to poke at over the holiday.

Web maps usually join together lots of small images called tiles (this is why you see square patches as google earth/map loads). They do this by querying a "tile server" API. It turns out this standard can also be leveraged to label and fine-tune models on map imagery. In my day job we built infra to efficiently serve imagery through tile servers for map visualization. So I wanted to test out ML applications of that infra.

If you've ever tried to use Google Analytics (GA4) and Google Search Console (GSC) to figure out what's working with your marketing, and what to do next to grow, you have probably got frustrated at some point.

It acts as a Marketing Strategist. You can ask questions like "why is my SEO traffic down this week" and it will give you a clear answer based on your site's performance data, as well as a checklist to improve.

https://refreshagent.com

It's intended to be anti-memetic, and anti-guilt trip. Just put it on your watch, install a program (open format) and you never need the phone itself. Your workout is a holiday from your phone.

The data can be exported if you want to use it elsewhere.

I originally made it for ROCKNIX but as there was no way to share the app I paid the Apple tax :/

https://github.com/jmahmood/RED-STAR-WEIGHTLIFTING

I've been doing a lot of assembly, C, WASM and plan to top it off with a look at GPU instructions and PTX. I haven't learned as much as in the last two months in years, it's been great. And surprisingly everything has turned out to be much simpler and easier to implement than expected once demystified.

Now to be fair, AI has sometimes given me pointers when I didn't fully understand something. Using Gemini 3 for free has been nice in that regard. However I consciously try to only implement code myself and to actually make sure I learned something that sticks.

Just made a landing page and then transfered its style to the app using Claude AI. Was so impressed that I paid for a supscription immediately.

Will polish the app and plan to launch next month.

its a web app where you make boxes, add images or text of what's in the box. then get a qr code that you can tape to the box and scan to see the text or images in the web app.

hoping to make it a lot easier to look for things in the storage unit. instead of removing all the totes and looking in them. Just scan and see if the description fits what I'm looking for

There don't seem to be many automated tools out there that fit my need for this, so building out my own solution I have complete control over makes sense. It's a lot of fun to build this out exactly as I want to, rather than trying to configure a bunch of tools that I'm not familiar with and that don't meet my needs exactly.

The tooling I'm building up around this should hopefully make it easier for myself to get my playlists and track ratings off of Plex if I ever decide to abandon it for music listening.

Basically a mix of Teardown voxel physics + Astroneer solar system setting + in a Valheim-like multiplayer survival game. We've been working on multiplayer voxel physics in Unity for years now, so its nice to finally have a product almost ready

https://weatherstage.com/

I had some custom build scripts and sites for my dad and myself and was thinking I could make a simple SaaS out of it. Super early and didn’t advertise anywhere yet since the actual dashboard is very simple right now but it works and I keep adding the features I want to use myself.

Example dashboard: https://warnitz.weatherstage.com/

If you want to try it out, I suggest you write me at hello at domain and I will get you going. Let me know the type of weather station you have!

https://bloomberry.com

And sort of in that same vein, I've been developing my own static site generator that I eventually want to move my blog to. It's almost certainly going to be a worse SSG than every alternative, but it'll be mine and that's worth something in itself.

Plus it's just been fun to make! I wrote some gnarly code to generate infinitely nestable layouts that I'm kind of proud of. It's the kind of code that's really cool but you can only code on a project for yourself, because if someone else had to debug it, they might say some pretty unkind things about you.

This allows library authors to do more, like defining webhook handlers and (simple) database operations. The idea is to move complexity from the library user to the author, making (API) integrations easier.

I think libraries being able to write to your database is a pretty powerful concept, and can enable a number of interesting use cases.

https://github.com/rejot-dev/fragno

Working on building an investment assistant backed by real time data. ChatGPT and Perplexity finance are amazing, but all of them are based on web search data only, which is a big limitation in finance since realtime data is important.

We have an agent that has access to almost every data point you can think of in the stock market (as much as we can get), which gets leveraged before answering.

And we also figured out ways to build amazing charts in between answer snippets, which looks very cool. Investors are usually very visual.

Parents set up accounts, kids log in with simple codes and work through curriculum-aligned Math and German exercises. Built with Elixir/Phoenix/Ash and LiveView.

The hard part isn't the tech - it's creating content that actually maps to the German school curriculum rather than generic "educational" fluff. Currently grinding through grade 2 math topics.

https://klugli.de

It can work already as a "Generic" ActivityPub server and it can be made to work with Client-to-Server API, but given that there are not mature clients for that, I am now in the middle of an exercise where I am taking the existing server and implementing Lemmy's and Mastodon's APIs based on top of it. Once I can get any Lemmy and a Mastodon client working, I will then start changing their own SDKs, and then I can replace calls from their application-specific APIs with direct calls to Linked Data server.

  [0] https://activitypub.mushroomlabs.com

It looks inside each file to see what it’s about, then moves it to the right folder for you.

Everything happens on your Mac, so nothing leaves your computer. No clouds, no servers.

It works in 50 languages (including English, German, French, Spanish, Swedish) and with images (OCR and object recognition), PDFs, Microsoft Office, ePubs, text, Markdown, and many other file types.

If you have messy folders anywhere on your Mac, Floxtop can help.

So, I'm building a toolkit that allows to keep things simple for the end user. Run Ollama and Open WebUI configured to work together: `harbor up ollama webui`. Don't like Ollama? Then `harbor up llamacpp webui`. There are 17 backends, 14 frontends and 50+ different satellite projects, config profiles that can be imported from a URL, tunnels, and a helper desktop app.

https://github.com/av/harbor?tab=readme-ov-file#what-can-har...

https://bsky.app/profile/zeta0134.bsky.social/post/3m7xuxuc3...

Currently mostly happy with where this has ended up, but the percussion is a tad too basic and needs more work. One thing at a time I suppose. :)

It's a work in progress, but it's at a stage where if you ask nicely I'll let you know where to download it.

There are a lot of apps that can be built on ATProto, the PDS, etc. If you are exploring the same space I'd especially like to hear from you. I'm easy to find, which is the most useful thing about being named Zigurd.

I’ve also been playing with Bun and I have a business idea that would be a good fit, and huge potential but I just don’t have enough time to start something new anymore.

A Python ORM, inspired by Drizzle and the like. Whenever I come to Python I'm frustrated by the ORM options. They generally lack type-safety on inputs and outputs, or useful type hints.

SQLAlchemy is an institution but I think it's hard to use if it's not your full-time job. I check the docs for every query. I want something simple for the 80-99% of cases, that lets you drop easily into raw SQL for the remaining %.

I'm going to keep hacking at it, would love to from anyone who thinks this is worthwhile (or not). Also: - The interface for update queries is clunky. Should I add codegen? - Should I try to implement a SQL diffing engine (for migrations). Or just vendor sqldef/similar...?

Recently it hit v2.0 spec conformance. 3.0 is next on the roadmap. (I'm executing it against the upstream spec test suite.)

I don't plan to make it a highly-performant decoder for use in production environments, but rather one that can be used for educational purposes, easy to read and/or debugging issues with modules. That's why I decided not to offer a streaming API, and why I'll be focusing on things like good errors, good code docs etc.

https://github.com/agis/wadec

P.S. I'm new to the language so any feedback is more than welcome.

https://news.ycombinator.com/item?id=46215686

https://github.com/mindflayer/togo

It can replicate a DB in as little as 9 seconds.

It's Open Core: Community Edition and Pro/Enterprise editions.

Still a WiP --> https://kopidev.com

The project has a CLI interface that is free and open-source, but you have to self-host the gallery. We are also building a SaaS app which is basically a managed version of the open-source tool with a visual builder and we take care of the hosting and CDN.

https://simple.photo

I’m still exploring new forms of AI-powered learning tools.

The latest thing I’ve been working on is an adaptive mode inspired by the LECTOR paper [1]. Where each lesson is a single learning concept with a mastery score tight to it based on your understanding of the said concept, so in principle the system can reintroduce concepts you didn’t fully grasp later on, ideally making separate flashcards unnecessary.

It can be self-hosted if any one want's to give it a try!

[1] https://arxiv.org/abs/2508.03275

It’s been incredibly rewarding to see people’s changing opinions of their local government

It plans multiple days ahead to make the best use of low prices and surplus solar.

It can use the vehicle api or the charger api to control charging.

https://akkuplan.eu

And I realized I couldn't give a concrete answer. Lots of speculation, but I realized I didn't have hardly any real data. Inspired by Adam Grant's work on "rethinking", I'm _currently_ writing a tiny CLI to run self-experiments on my own productivity, auto-checking in / observing commits/code changes.

Goal at the end is to be able to test myself across different dimensions with "no AI", "moderate AI" (e.g. searching, inline assist), and "full AI" (agents, etc). https://github.com/wellwright-labs/pulse

TLDR the incremental compiler rewrite is finally bearing fruit. Namely, because we no longer have a batch compiler (i.e. we don't bail on the first error), we can

- provide LSP results (hover, goto def, etc) on non-broken parts of your isograph literals, even in the presence of errors

- surface those errors in VSCode, and

- fix those errors with auto-fixes!! (https://www.youtube.com/watch?v=6tNWbVOjpQw&t=314s) Which is to say, select a field that doesn't exist, and let the compiler create the isograph literal declaring it.

It's a great feeling to see this level of DevEx

I wrote about it here: https://pcmaffey.com/custom-ssg/

Forkable template: https://github.com/pcmaffey/bun-ssg

Killer feature is multiple plans per customer.

https://railsbilling.com

https://bloks.run/

I wanted something local and offline first + 10-20% better than excel, think I'm missing a few features other might find useful, but it works for my needs which has been great.

The vast majority of tokens in a sequence will be irrelevant to an attention mechanism outside of a very small window. Right now however we tend to either keep all cache values forever, or dump them all once they hit a certain age.

My theory is that you can train model to look at the key vectors and from that information alone work out how long to keep a the token in the cache for. Results so far look promising and it’s easy to add after the fact without retraining the core model itself.

https://typequicker.com

This is something that started as a passion project - I wanted to see just how effective of a typing application I could make to help people improve typing speed quickly.

It’s very data driven and personalized. We analyze a lot of key weak points about a user’s typing and generate natural text (using LLMs) that target multiple key weak points at once.

Additionally we have a lot of typing modes.

- Code typing practice; we support 20+ programming languages - daily typing test - target practice; click on on any stat in the results and we generate natural text that uses a lot of that (bigrams, trigrams, words, fingers, etc).

https://www.yups.io

It is very stupid for now but I am working on the process and a friend of mine is working to improve the LLM (that's the project Babelfish).

Developing Qt apps with C++ and QML is a blast - the fast performance of C++ and ease of use of writing UI in QML. But there is so much left to be desired with the built-in Qt Quick components - mobile issues like non native text handling, non native swipe-able stack view and much more. I’m aiming to bridge that gap.

[0] https://apps.apple.com/us/app/reflect-track-anything/id64638...

Working heavily right now on Customer Personas to use in validating/invalidating , which are configured with viewpoints, biases, and tendencies. Coming very soon will be Persona Journeys, in which you can get live, goal-oriented evaluation of your web app by a Persona.

Using an esp32, high speed ADC and 4 bass guitar pickups to detect and reverse engineer the club's path and face angle as it swings past the pickups.

This is the PR: https://github.com/ironcalc/IronCalc/pull/616

Feel free to comment and destroy it!

You can test it in: https://testing.ironcalc.com

A lot of teams enjoy using Linear for product management but still have to use Notion and Confluence for knowledge management. I’ve built Outcrop from the ground up to be fast with much more reliable search and realtime collaboration.

Hundreds of teams from startups and major companies have signed up for early access and many have made early commitments to support the development of Outcrop.

If your team would be interested, I’d like to hear from you!

https://outcrop.app

https://imedadel.com/outcrop

imed at outcrop.app

We have an ML model that's trained on real reservations and use an LLM to decide why a user mightve opted out. We apply personas to this LLM to get a bit of a sense how they would probably be operating the booking flow.

I am trying to offload as much of the complex stuff to existing parts of the kernel, like using systemd/cgroups for resource limiting and UNIX sockets for authentication.

Repo should work with any github hosted changelog file. https://github.com/stevenmenke/claude-code-changelog-rss

Was hoping to have these ready for Christmas season, but life as always gets in the way!

-> https://next.nocodefunctions.com

A complete refactor and stack change so that the web app can be more easily extended to new functions.

Opinionated workflows and automations for less technical teams where no code, low code or vibe code tools are beyond reach.

https://housecat.com/

https://hortus.dev/s/badges

Currently in the works are a digital sand timer which can be used to track pomodoros (or any sequence of time intervals), and a Jovian orrery which displays the positions of Jupiter’s moons on a strip of addressable LEDs.

This has been a fun project so far for me:

* First time using Claude Code. CC has made writing code fun again (I'm an experienced software developer, with - gasp - over 20 years of professional experience).

* On macOS, WhisperKit + Apple Intelligence (SpeechAnalyzer) is a powerful combination for offline transcription.

If you're interested in joining the beta, feel free to send me an email: diarmuid.glynn@gmail.com. The software is working now, but the documentation and website ( https://www.algomommy.com/ ) are unfinished, so I'd like to provide direct support to any interested beta users.

Build to help you save and organize links without friction. Group related content into collections, pin critical resources for quick access, and search your entire knowledge base instantly.

Right now I am tinkering with wails (https://github.com/wailsapp/wails) to build an app store.

https://news.ycombinator.com/item?id=46097671

https://github.com/blue-monads/potatoverse

Ai-rganize — For using AI to sort files/folders on your local environment (Mac, Windows or Linux). (https://github.com/adefemi171/ai-rganize)

yaml2mcp — Got tired of writing MCP servers in JSON so I decided to build this as well. (https://github.com/adefemi171/yaml2mcp)

It's very unstable at the moment but plan to have it fully implemented and working by the end of next month.

Using it to build a virtualized computational storage device for research.

:)

https://github.com/Chrilleweb/dotenv-diff

https://dhuan.github.io/mock/latest/examples.html

Planning on wrapping up the year with a year in review post (thankfully I've been writing monthly updates as I go, should save some time).

Apart from that, clearing up tech debt that helped me ship fast, but was ultimately a bad fit for the business (Next.js and GraphQL).

Eventually I'll open source it, but I'm a bit shy so I want to open source it once it's done without a commit history.

https://marketdao.dev

My initial goal is to make a functional SillyTavern (AI roleplaying) replacement. SillyTavern builds prompts from a few rigid buckets (character, scenario, lore, system prompt, author's note...), which makes complex setups hard to manage. Content gets duplicated, settings have to be toggled in multiple places, and it’s easy to accidentally carry or modify state across conversations. Over time, it becomes difficult to tell what context is actually in effect.

I’m building an alternative that treats context as small, reusable pieces that can be composed and organized flexibly, rather than locked into fixed categories. Characters, settings, and behaviors can be mixed, reused, or temporarily enabled without duplication or manual cleanup, and edits preserve clear history instead of rewriting the past. The goal is to make managing complex context deliberate and controlled instead of fragile.

Although I’m trying to get the functionality required for roleplaying done first, the app is generic enough for other AI workflows where fine-grained, explicit context control is an improvement over existing chat interfaces. Think: start a new conversation with an assistant and start checking off rules, documents, and instructions to apply to the chat. Regenerate responses with clarifications or additional one-time context layers.

A local, cli based task and record manager, focused on simplicity and speed but includes support like managing schedules and records and searches etc to support it being a structured schedule helper.

It’s not a trading tool or product. More like a weekly, machine-assisted research project. Each cycle I run analyses on 120+ public companies across semiconductors, cloud, biotech, energy, robotics, quantum and crypto. The framing is inspired by Emad Mostaque’s “The Last Economy” thesis — the idea that when intelligence becomes cheap, the physics of value creation start to look very different. I originally built it for myself and retail investors in my family but I figure it could have more general utility so prettied it up a bit.

The system uses large-model reasoning (GPT-5+ though I've also tested Sonnet, Gemini and Grok) combined with structured scoring across technology maturity, risk, competitive positioning, and alignment to AI-era dynamics. The output is static HTML dashboards, PDFs, and CSVs that track month-over-month shifts. I'm adding to it weekly.

Mostly I’m trying to answer questions like:

* Which companies are structurally positioned for outsized upside in The Last Economy?

* How should I deliver the research so that it would have been actionable to someone like me 30 years ago?

* What signals would help folks identify “the next NVIDIA” 5 years earlier?

The inference costs real $$$ so I've set up a Patreon that, hopefully, will allow me to scale coverage and extend the modelling and methodology. There is a free tier and some recent, complete example output on the web site. I'm also happy to gift a free month for folks willing to provide constructive feedback: https://www.patreon.com/NextArcResearch/redeem/CC2A2 - in particular I'm looking for feedback on how to make the research more actionable without drifting into "financial advice".

I don't collect any data but Patreon does for authentication and Cloudflare does to deliver Pages. The Last Economy is here: https://ii.inc/web/the-last-economy

Not sure if I'm missing a better tool but trying to keep a good working mental model of this has been a nightmare for the operators I've maintained.

[1]: https://apps.apple.com/de/app/days-of-life-milestones/id6738...

A citizen service initiative that aims to serve as a platform for monitoring areas of need in Puerto Rico.

- Added creating blog posts

- Improved moderation tools

- Rewrote an upstream client to move off deprecated API

- Lots of improvements around CSS/ui (many thanks to Gemini)

- Fixing lots of bugs

Longer term personal aim is a self-hosting platform based on k8s with straight forward bootstrap, similar to Yunohost but k8s based.

Attracting new monthly sponsors and people willing to buy me the occasional pizza with my crappy HTML skills.

https://brynet.ca/wallofpizza.html

Started this out by building a spreadsheet controlled by an LLM. Now putting a direct filesystem inside, simplified enough to have programmatic control of slide builders, spreadsheets, terminals and vibecoding applications

While trying to figure out a good ICP and reach PMF

and also I am trying to find some time to improve the minimal month planner https://printcalendar.top/

https://www.puzzleship.com/

In the time-honored hacker tradition of added more problems to the problem i'm trying to solve I'm learning a new language (never done FP before, either), building the product I wanted, using the latest crop of creative tools, and treating it as a little end-to-end business startup too. Launching in January!

While I was working on the tablet interface (in Godot Engine) I put Claude to work on what after two minutes became a full product on its own with a new file format as well. Tell me what you think! (so far the response is meh...)

https://habitatview.app

https://omnispect.dev/

I was pondering doing something in regards to decentralised consummation of content. I am beginning to see how various websites are walling off their content and centralising everything whilst also monetising access to it for themselves and kicking content creators out, forcing them to run their own websites and use multiple backup platforms(mostly the dying youtube).

So I was thinking about flipping it on its head and instead of going to different websites to consume this content, like youtube, twitter and whatnot, people would have a single program to aggregate it instead. Then it occurred to me that this is what RSS/Atom was made for, kind of. So I am just letting the idea marinate for a bit and maybe next year I will look into it. Mastodon might have some good concepts in it that I want to look into and also come up with some standardised way for richer content that creators could provide beyond RSS to make it more palatable and easier consumable for users.

tl;dr not much this month :)

Buying, researching and analyzing automotive data is broken. Trying to fix that bit by bit

https://github.com/tirrenotechnologies/tirreno

• I open-sourced and released some iOS dev tooling I built for Claude Code that multiplied my personal coding productivity: https://news.ycombinator.com/item?id=46264591 Nobody cares yet, but it makes me feel good to share something cool.

Also if anyone needs a contractor hmu at https://elephtandandrope.com

Also working on youtube vids to teach people to code for personal branding and another channel for POV driving vlogs but editing eats time :(

Just whatever time can allow really!

Also, getting into the guts of how agents work and messing around with the knobs and levers is super interesting and where the real differentiating skills are

Built on ADK, CUE, and Dagger

https://github.com/hofstadter-io/hof/tree/_next/lib/agent

(my swiss army knife for dev work, getting a rename soon(tm))

Check the fireproof video, it's quite fun haha https://www.youtube.com/watch?v=v0NXXfCA2CY

Vine but for user-submitted microgames

Docs: https://xelly-games.github.io/docs/intro

Just finished a major (v0.10) revamp of the API (you can use connet as part of an application, not through the CLI) which also fixed a few issues I've been seeing before.

Now, I'm gearing to update the relay protocols - currently relays are closed off by the control server (e.g. you ask it to provision you a relay resource) which requires the relay to communicate with the control server itself. In the new version, the relays will be operating on their own (there might be a shared secret with the control server, in case you want a closed off relay) and peers will reserve directly with the desired relays. Maybe in future, the relays might form clusters on their own to take advantage of better relay-to-relay network and peers will reserve only at the relay closest to them.

Another stream of work, is giving peers identities. Right now the server will give them an internal identity to better support reconnects, but these are not stable (e.g. they don't survive client restarts). In future, the peer will advertise their identity and then other peers may choose what peers to allow comms with and what to ignore, pushing more decisions into peers themself.

Yet another change I'm thinking about is exposing raw endpoints to enable users of the system to implements other protocols - I'm not quite sure if this is really needed (the destination/source, e.g. server/client) covers a lot of ground by itself, but it would be great if these are not the only options.

Many options how to continue, but if I'm out of ideas, there is always a Rust rewrite to throw in /s