Ask HN: What are you working on? (May 2026)

We're working on Drawers (https://drawers.computer), a macOS app to give each of your projects its own dock, space, and windows.

We integrate with macOS spaces to switch out a project-specific dock on each space, containing only the resources you need for that project. We made it possible to add granular resources instead of full apps to the dock (think specific slack channels instead of the whole slack app), to keep the dock hyper focused on what you need.

We built this to stay focused while working on the computer, and we thought that the native interface mixed all our projects together, causing us to get distracted.

Looking for beta testers! Free download from https://drawers.computer

I’m continuing to work on my daily puzzle game Tiled Words!

https://tiledwords.com

Forbes just wrote an article about it which was a fun surprise! [1]

It recently turned 6 months old which is wild to me. My wife and I have made a new puzzle every day for half a year! I wrote a blog post about this [2]

I recently released user logins. That went well and a lot of people are using them. I also let you filter the backlog by completed puzzles based on player feedback.

This week I’m going to start releasing player submitted puzzles and release my puzzle building tools. You can watch a video for a sneak peek of those tools. [3]

1. https://www.forbes.com/sites/barrycollins/2026/05/02/bored-o...

2. https://paulmakeswebsites.com/writing/six-months-of-tiled-wo...

3. https://m.youtube.com/watch?v=d8_zhMKd0Yg

European night trains search engine. The plan is not to provide the functionality where user searches for a specific route on a specific date and the engine returns prices. The plan is to provide "tips for trips", e.g. I provide a starting city and it will recommend me interesting trip ideas any time in the future. There are many flight apps that provide this functionality, but no train app.

Currently I'm building scrapers for all relevant provider, then I want to connect the data for multi-city trips recommendation. Plus some connection to the day trains so that the trips are built more easily.

Some existing trip recommendations:

Prague <-> Amsterdam https://trainbot.eu/?from=prague&to=amsterdam&type=return

Berlin <-> Paris https://trainbot.eu/?from=berlin&to=paris&type=return

Zurich <-> Budapest https://trainbot.eu/?from=zurich&to=budapest&type=return

* This is my first time actually posting anything on HN.

I've been making a DSL for writing sheet music specifically for drums as raw text, inspired by ABC Notation (but of course just for drums).

Now writing this I noticed that it's kind of complicated to explain and having a landing page would make my life so much easier.

But the gist of it is, you write notation that looks like this: https://gist.github.com/Luigi123/945af7e5cc8dfbfd186f0a99754... and it renders sheet music in PDF, and also allows you to play the same music as a game (DrumMania / DTXMania style).

Now the language / compiler itself has been working quite well and I've been dogfooding it for like six months now. The next thing is an IDE-style editor where you can import a song and write the notation following it. Making THAT has been quite the journey. Here's a screenshot for good measure: https://i.imgur.com/EmlqlrM.png

Acoustic diagnosis of electrical problems on the electric grid!

I'm building a tool that allows you to determine the health of an electric transformer from only your phone. It tells you:

    - the loading
    - the health of the windings and core
    - and whether the phases are unbalanced

I used to be a submariner, so my professional background is in power plants and sonar analysis, so I'm getting to combine the two in this.

Acoustic diagnosis of electric issues is FASCINATING, and it feels like there hasn't been a lot of research into this, so I have been slowly chasing down various acoustic patterns I find and try to derive them from first principles of physics.

I'm making an iPhone app for it, and Xcode has been truly awful: non-deterministic, crashing all the time, and error messages that tell me absolutely nothing. I would like to use xtool, but it doesn't have the preview, which I need for debugging.

I finally managed to finish a project and publish my first game on Steam: https://store.steampowered.com/app/4195030/balls/

It's a short chain-reaction game in which you explode balls bouncing in the screen, and need to build up to target scores. You build bigger and bigger combos as the game progresses.

It was a blast to work on it, starting with a small toy and just adding features that "felt right" until I had a game that was fun to play. It was quite hard to find a balance though, so a lot of numbers are arbitrary - but I enjoy seeing people breaking the game in new ways and finding new builds.

These days I've been working on patching reported bugs and sharing the game with people. Now after the latest patch, I feel like I'm done, but I feel like going back at it and adding an idle mode. And maybe simplify the codebase so I can test and iterate better, and then add many more ball types...

I know that any good LLM could replicate this pretty quickly, but I made this myself and I'm still feeling proud of the accomplishment :)

I'm constantly working on https://only-eu.eu/en

A directory of European software and general alternatives to popular products.

Think cloud storage, email, VPN, browsers, smartphones, bikes,... About 175 products across 30+ categories right now.

Next categories will be: personal health, commercial e-mail and newsletter management.

If you have a product that's missing here, please feel free to suggest it via the suggestion form on the site.

[NO-AI]

Being a weightlifter for 20+ years now, I'm working on a barbell speed and path tracking sensor based on newer IMU hardware technologies, which makes it both more precise and cheaper than camera- or actuator-based systems. Ultimately it helps you lift and train safer and better.

It's an intersection of industrial design, hardware, firmware, and software (and some sport science, of course). This intersection is not yet dominated by LLMs so it's a breath of fresh air.

In an early prototype stage as in "strap a Raspberry Pi to a bar", but it looks promising and I'm happy to move forward, also using connections from my previous 12+ years in China.

I’m an application developer by day, but lately Claude Code and Codex have finally made microcontrollers approachable enough for me to start tinkering with them on the side. I built this little “holographic” display that shows the surf forecast for any beach. While my friend built the casing, and mechanical part of it

https://x.com/paulnovacovici/status/2041722840190480581?s=46...

A small game called Colorhunt to play with my friends https://colorhunt.lol

You get assigned a random colour and have 24 hours to go out and take photos matching it. The game then generates a photo grid from everything you captured.

Modes:

- Solo → 9 photos by yourself - 1v1 → compete against a friend, combined grid at the end - Squad → everyone contributes to a 20-photo grid

No accounts, no app install, no personal data stored. Photos and generated grids auto-delete after 24 hours from Cloudflare R2 storage.

Made it for fun to find a way to do a shared a activity with my girlfriend and also to challenge my friends over the weekends.

Currently redesigning the frontend flow but I am kinda poor at design

Would genuinely appreciate feedback, ideas or anything

The last couple years I've been prototyping a not-for-profit pinball museum in Chicago. In the coming weeks we'll be opening 7 days a week in a 2900 sq ft space in the Loop: https://theflip.museum/

It's my first time starting a physical, retail business and it has been quite an education in the small details.

Working on fmsg - open protocol for instant messaging, distributed by domain like email..

Its a message definition and protocol, addresses look like @user@domain, anyone can run a host, and threaded messages are linked by cryptographic parent hashes..

The idea is to take the best from email: open protocol, domain ownership, interoperability (unsolicited mail is a feature not a bug), and the best from closed instant messaging rebuilt: efficient binary messages, conversational threads, sender verification, message integrity etc. built-in. Originally envisaged for human-to-human messaging but partculalrly interesting time right now with human-to-agent and agent-to-agent messaging...

The OSS stack is up and running: Go host, Dockerised full setup, CLI, Web API, and a spec nearing v1.0. Did Show HN post week ago: https://markmnl.github.io/fmsg/show-hn.html

Seeking feedback, criticism, validation :) protocol bikeshedding, and especially interest from founding-engineer types who want to help build an open messaging ecosystem rather than another closed app..

This starts from my frustration when opening large CSV files, but later evolved to a log/data analyzer that loads arbitrary format in constant time (O(1)).

https://github.com/Verticalysis/Hitomi

The secret: I engineered an incremental combinatorial parser capable of processing customized format from a steam. Any inputs, including file or the stdout from a command, are first chunked and then fed to the pipeline. The UI is ready when the first small chunk is processed.

Other highlights: 2-mode filter, one with a convenient UI and the other is based on an extensible DSL for complex cases;

Timeline mode scrollbar, a secret weapon for log or time series analysis;

Column widths fit to content automatically;

Native code, no web bloat;

Cross-platform (currently Windows and Linux, MacOS WIP).

If you are tired of all the quirks Excel have when working with CSV files, you'll gonna love it!

I’m developing X/D Loom (https://xdloom.com), a tool that helps car enthusiasts create automotive wiring diagrams.

About a year ago, I engine-swapped my Nissan D21 hardbody from the Z24 petrol to a TD27T turbo diesel and also installed a whole bunch of accessories, like spotlights, a winch, and an air compressor. But being lazy, I didn’t write down any of the wiring changes I made while doing all of this. So fast forward a year, and now I can’t remember how all the wiring works.

My current project car is a Jeep Cherokee FSJ, and for it, I want to build a completely new loom from the ground up. So to try and avoid making the same mistake I made with the Nissan, I Googled “create automotive wiring diagram”, but all the results were for complex enterprise grade solutions charging $200/month. That’s why I created X/D Loom as a project car guys' tool for creating wiring loom diagrams. It allows you to drag different electrical components onto a canvas, connect them with wires, and export them to a PDF or PNG.

I've shared this one before but I built a logistics management system to power deliveries for a business I founded years back and I've continued to refine it since:

https://toanoa.com/

Since the initial MVP, it's done close to 100k orders and I've added new functionality like:

- Intelligent order batching & route optimization that can interleave tasks across orders in such a way that they still have the best chance possible of completion within their delivery windows

- Further refined the mobile tracking logic in our driver app to improve the quality/frequency of position updates while continuing to be as efficient as possible on battery

- Numerous backend/DB optimizations such that average response times are in the tens of ms at the current volumes it's handling.

It's not open source but if you have an interesting use case and are curious about it, feel free to reach out.

I've been working on cardcast.gg. It gives you the ability to play Magic: The Gathering with your friends remotely using a webcam.

I got back into MTG back during the pandemic after a long hiatus and Spelltable is what brought me back. My playgroup lamented more features and something tailored to our needs, so curiosity got the better of me and here we are. :)

I've never worked with computer vision before, but I went through a whole journey that started with the classical computer vision techniques and ended with recently migrating to the transformer-based models. Been a really cool adventure!

My playgroup has been consistently preferring it over Spelltable and have been wanting more and more features. I would love for people to try it out and start building a community around it! Discord is on the site.

https://cardcast.gg

After quite a few years of coming up with and implementing 'great ideas' but not being able to follow through to making them revenue generating products, I'm on my best bet so far.

I always wanted to build a real-life puzzle game, which is app/mobile assisted. Had yet another eureka moment, and built a usable prototype (backend plus iOS app). Good feedback from a small circle.

For a while I was aware of someone (I knew by sight) who worked in the same sort of subject matter (but a non-tech). I approached her, we had a coffee, I pitched the idea and how she could bring it to life, as I made the tech side. She jumped on board.

We're two and a half weeks in, have gone full speed and are making something great (for our audience). My future co-founder is amazing, great insights, opinions, drive. We're potentially launching in a couple of weeks, a free/MVP version of a puzzle game.

I've been through many iterations of trying to get something off the ground. Tried tech co-founders, and the last years of going solo (very hard after you've done the coding). But this now feels right. A puzzle app/game for every day people to have some fun. And a future co-founder whose life is outside tech, who's bring a sort of fun energy outwith let's make loads of money or isn't the framework/AI cool.

Balance is good. Contact with reality is good too :)

Working on a tactical map-based WW2 submarine simulator called Silent Shark. https://silentshark.app

Free beta version is running well ( https://silentshark.app/alpha ) and I plan on releasing the full WW2 campaign version on Steam, App Store, and Play Store in the next month or two.

It's been an absolute blast getting feedback from Navy geeks on Discord, tweaking things, and my favorite moment was when my stadimeter instrument (finds distance based on angle + mast height in the periscope) worked without any "cheating" on my side simply because math works.

I’m building a blackjack card counting tool for people to learn how to count and how to identify games that are winnable. It is designed to take a completely novice to an advanced, winning card counter, using a Duolingo like approach - mastery based learning across sequential modules. Minus the ads and dark patterns.

I've been working on RVW, my adaptation of the standard transformer model that is capable of online continual learning without catastrophic forgetting. I finally published the first pre-print of my early experiments: https://doi.org/10.5281/zenodo.20064617

Now I'm working on expanding the work into more parameters and improving performance. I just finished an extremely harsh test of a Nemotron-flavored RVW that consisted of stretches of a random assortment of domains interspersed with long runs of single domains. Across all of it the model didn't forget (and actually improved on some of the more challenging domains). PPL on SmolTalk is still in the ~18 range, which I'd like to get lower, but this is all with only 4B params.

Currently, I'm training a Llama 3.2-flavored RVW with only about 2B params to see how that turns out. Depending on results of that, I may take it to Gemma 4 next.

I’m smelting ore!

I got into creating my own rings, and I’d really like to create one with ore I harvest myself. Gold is too hard and silver can be kinda dangerous, but malachite is pretty safe and I can just drive to Copperopolis to pick some up.

Basically: smelt the malachite with flux and charcoal to get pure copper, flow that into an ingot mold, hammer it into shape. Then I’ll have my own ring, with metal I collected with my own hands

A better permissions layer for coding agents. The tool works like auto-mode for Claude Code, so you can stay in the flow and only get prompted to allow or deny tool calls when it truly matters, but it is fully deterministic. My benchmarks surfaced that most Bash calls don’t need an LLM to be classified as safe, ambiguous, or dangerous. A deterministic classifier can auto-allow or block 95% of Bash tool calls as safe or dangerous, with only the remaining 5% being truly ambiguous or unknown.

Conclusion is permission reviews with LLMs like Claude’s auto mode or Codex auto review are like using a data center to flip a light switch - overkill.

The main benefit is that your agent’s autonomy can be governed deterministically through policies that can be stored at the user and repo level. The bonus is that you save tokens vs using auto modes.

https://nah.build

I left my fulltime faculty position at the university. I'm only teaching two modules this semester and I'll probably fade out even more in the future.

I've also paused recording any new lectures at pikuma.com for now. I'm still taking some time to decide what's next for the website. I'm currently focusing on reviewing math & physics to help homeschool my son.

Other than that, I just improved the roof of my chicken coop and I'm slowly evolving the foundation of my study cabin. This new place will be my offline library and music room soon. My plan is to only really go online once a week starting August this year.

I'm continuing work on my daily procedural logic puzzle, Ruly.

https://playruly.com.

You play by setting rules onto a small grid of numbers to maximise your score.

My focus the past few weeks has been on refining the difficulty by experimenting with different rule types, and improving the UI.

I'm pretty happy with the look and feel now but feedback is always welcome, and I'm especially keen to hear what you think of the level of difficulty of the puzzles. It's a tricky balance to introduce variety without adding complexity.

There's a (very) small contingent of daily players now which is really motivating.

Your comments very welcome.

I'm working on https://www.OnetimeFax.com - it's a way to send and receive faxes for a flat fee without needing to subscribe to anything.

Version 1 was a script I wrote for myself to fax the courts a jury duty deferral during covid lockdown, and then during my parental leave I productized it.

I've been learning to crochet. I'm trying to do more hobbies with my hands, but it's also pretty interesting from a mathematical perspective. The fundamental primitive (the chain stitch) is like a series of slip knots, and each stitch is reversible. So the piece is actually a series of reversible transformations. The yarn is sewn in at the end to secure it.

This has some interesting implications. If you make a mistake, you can always backtrack and try again. If you have a crocheted piece, at least in principle you could find the lose end, free it, and work back stitch by stitch to reverse engineer it. (In practice people don't seem to do a stitch-for-stitch reverse engineering just like you probably wouldn't bother reimplementing something line by line without a compelling reason, you figure out what's going on in the challenging places just by look and feel and improvise from there.)

I'm oversimplifying somewhat and there are some forms of crochet that include irreversible stitches, yarn can be felted together (entangled, like a cotton ball) to create irreversible bonds between adjacent strands, and often several panels/pieces are joined together irreversibly to create a larger piece.

I'm working on a diffusion-powered UI design tool. My short term goal is to make AI-designed UI not look like Tailwind. My long-term goal is to be Figma, but powered by diffusion.

https://diffui.ai

I quit Figma about 4mo ago to start working on this, and the gpt-image-2 drop really legitized the bet. I recently release Brands for diffui, which let you establish a design system and consistently generate with it. I made a Brand out of the recent UFO files release, which allow for some really fun designs:

https://diffui.ai/brand/2ff1b00a-d698-43ea-a42e-7c4a2e670c04 (no account required to generate with this if you want to try)

I made a Python tool to build distroless container images for projects managed by uv. It draws inspiration from Ko from the Go ecosystem and works with/depends on uv from the Python ecosystem, so I smashed them together and called it Kuvo: https://github.com/hxtk/kuvo

It’s a hobby project in a very early state where it technically works but it’s missing several things I think it needs before I’d use it for anything serious. As of right now it isn’t even complete enough to dogfood a minimal container for itself without an intermediate base image because it can’t target a platform compatible with the distroless uv container image.

Still chugging away at my NES rhythm game. Currently, in addition to climbing the content mountain (so, SO much pixel art and music needs to be made) I'm also slowly learning video editing workflows. I was able to put together a brief gameplay trailer this last week:

https://store.steampowered.com/app/4129270/Tactus/

Right this second I'm looking for an alternative to After Effects that runs on Linux systems, as kdenlive has some limitations with its layering implementation. I'll probably give Blender and Godot both a whirl, as I want to get more comfortable with those tools for future projects.

I've just published my first book. The central argument is simple: software development is a design activity, not a construction activity — and confusing the two is the root cause of most project failures. Written for developers, managers, and anyone who has ever wondered why building software is so much harder than it looks.

It's on Amazon in both Kindle and paperback formats.

https://www.amazon.com.au/Code-Design-software-projects-deve...

Research grade orbital mechanics, specifically of asteroids/comets. I've been working on it for 4 years now, finally tried using some AI tooling the last few months and ended up vibe coding a fun little visualization.

(Desktop Strongly recommended) https://dahlend.github.io/ketev/

I'm working on a general repo shape/structure linter (language agnostic)[0] - the idea is to enforce things like directory structure, existence of various files (LICENCE, etc.), file naming patterns, jsonpath + schema over json/yaml/toml, absence of potentially malicious unicode. It comes with rule bundles for various languages/presets which can be combined and extended. A goal is for it to be very fast, and useable on huge monorepos. I noticed myself having to add various forms of validation/scripts when coding using AI, and decided to build a reusable, fast tool for this purpose instead of rolling validation scripts for each project.

[0] https://github.com/asamarts/alint

I am currently building a powerful TUI database client (that supports Turso as well) as I grew fed up with heavy db clients that take gigabytes of ram when rendering data that is text. It's also very convenient since you can launch it with a saved connection from any terminal anywhere.

It's still early in 0.1v at https://github.com/okira-e/TUIQL but already supports everything day-to-day that you would need.

[Vaava](https://www.vaava.app/) is a baby tracking/logging app I originally built for myself, now available on both app stores. All the user generated data is stored only on device and is transferred in local network to users who you have paired the app with. There is 0 behavioural analytics, even the crashlytics are 100% optional.

There is a couple of semi-unique features; you can use your voice to dictate and generate events (feeding, sleep etc), you can also scan documents for growth measurements.

You don't need user account to use it, there is no subscription, the paid features are available behind a single purchase for lifetime. Still like 90% of the features are available for free.

Also https://www.athilio.com/ , which originally was also purely for my own use. Most sports and fitness wearable manufacturers own software and 3rd party software make it incredibly hard to do "how does my this months metric x compare to same month last year", athilio attempts to make those queries easier. Many of the ideas are basically copied from software observability concepts. Also I have used the app to implement and learn agentic workflows.

I'm building Deptheos (https://github.com/ryanmitts/deptheos), a photography/macro focus stacking program.

It works on MacOS, built with Swift and Metal. My goal is to make a super fast, and free, focus stacking program. I provided a notarized MacOS DMG for the initial release, but if built yourself, it will run on an M4/M5 series iPad Pro as well.

The core ability I wanted was to support RAW files as inputs, with DNG files as outputs. This is done using either LibRaw, or Adobe DNG Converter (runtime options).

I have been really into macro photography the last couple years, and have been slowly working on trying to build my own program to handle the focus stacking.

I'm building https://clipper.dev

Docker is...quite slow with large images. I've built a registry+pull client+buildkit builder to make it better. It splits apart layers, allowing for files to be shared between related images. In a robotics context, it can make pulls 10x faster. And in a cloud context, the format allows for pulling an image in 15 or 20 seconds instead of 60, without having to do a FUSE w/lazy pulling. Builds are faster, I store 7x less data due to better deduplication, I can run security scans faster due to not having to unpack tarball layers, etc, etc. I want to be the default registry for all ML related work, in the future.

I do calisthenics 3×/week plus Ironman 70.3 prep, which means my training lives across Garmin, Polar, Withings + FIT files and front-lever sessions that no mainstream app models. So I built one that does both (and have been using for the past 4 years+): logs custom strength moves (front lever, FLAC, ¾ pull-ups), aggregates the connected devices (Polar, Garmin, Suunto, Withings, Apple Health) into one weekly view. Currently trying to see if can integrate some AI insights to my training routines. App is free for now as it does not cost me much (only servers for now), comment / use cases welcome: https://obitrain.com/

I built on online multiplayer boggle game back in 2008 that somehow drew a lot of users, many of whom still play every day after 17 years. About a year ago I started a rewrite from scratch in more modern technologies but stalled out after getting to about 80% of the way there. A few months ago Claude enabled me to finish the remaining 20% and was able to relaunch mostly successfully! It's been tough though. I'm a dad with three kids and use Claude all day at my day job and my interest in working late isnt always there. But I'm eeking my way to something that hopefully can stay up for another 17 years.

https://serpentinegame.com

Opal: Browser first feature-rich markdown editor, workspace and publisher

- https://opaledx.com - https://github.com/rbbydotdev/opal

No logins or sign-ups, totally free, MIT Open Source

Integrates with git and github, publishes to aws, github, netlify, vercel, cloudflare, great drag and drop image integration

I'm working on https://engine.build

It's a durable orchestration system for AI code generation which solves the problem of not being able to trust LLMs to complete long running (and high quality) implementations without having to babysit them and monitor the process, which is what I think is the most exhausting part of coding with AI.

You start with a spec or programmatic task list and the engine runs the whole workflow: implementation, verification, review, fixes, and finalization.

It treats agentic coding like a durable CI-style process, with state, retries, reviewer feedback, commits, and auditability built in. It's externally orchestrated, meaning it's not the agent running the loop, it's simply agents being used as tools and spawned in the loop as needed without awareness of the loop itself.

It's going to be open sourced soon and it's not meant to replace your IDE or Agentic Harness of choice. You keep using codex/claude code/open code/cursor/pi whatever you want and simply delegate the actual implementation to the engine, through MCP/CLI and other integration points.

It supports any LLM provider so you can have GPT 5.5 implementing and a mix of Opus 4.7 / Deepseek v4 Pro / GPT 5.5 reviewing at every phase for example.

Sign up on the website or follow us on https://x.com/enginedotbuild or me personally on https://x.com/aljosa , desperately need more followers :D