I'm building a very casual daily price-guessing game for my mum. Every day she gives me feedback, and I'm using it as a chance to de-rust my CSS/React + see how daily games tick.

https://scandle.co.za

I made an idle version of the 1999 MMORPG "EverQuest". There's maybe around 50 people playing at any given time and has a enthusiastic discord group for it. It's relatively fully-featured to the original game, and has a lot of new mechanics to make the idle format work well. The 3D graphics aspect of it is really more of a screensaver, though, and all game interactions are done through menus.

I recently converted a bunch of stuff to be client side instead of server side (turns out running a real-time MMORPG server is expensive) so there's a new round of bugs I'm still resolving, but it's still fun to play:

https://www.idlequest.net/

Building a self-hosted agentic OS I call AEGIS — Adaptive Execution & Generative Intelligence System. Running on a single workstation with a consumer GPU. The core idea is a three-tier model cascade: a cloud model handles architecture and review, a local 32B model handles execution and code generation, smaller local models handle evaluation. The cloud model never executes directly — it reviews diffs and approves before anything gets committed. The interesting problems so far: GPU arbitration across competing inference services using a distributed lock, giving local models read-only access to institutional memory before task execution so they're not flying blind, and autonomous fleet provisioning — I spun up a new server node last night without touching it after the USB went in. Next phase is adding department queues so the system understands context — infrastructure work vs. client consulting work vs. internal tooling — and idle-time priority advisory so it starts anticipating what I need rather than waiting to be asked. Goal is something closer to Jarvis than a chatbot. Early days but the bones are solid.

Experimenting with: https://github.com/chocks/locode Local-first AI coding CLI. Routes simple tasks to a local LLM and complex tasks to Claude.

also just posted a Show HN thread: https://news.ycombinator.com/item?id=47310543

Recently two selfhosted open source file sharing platforms archived their repos indicating no further updates (most recently https://github.com/kyantech/Palmr).

I wanted an opportunity to learn more Svelte so I created Enlace which has a Go backend and Svelte frontend.

It's nearly complete but I would love any assistance with testing.

Adding new openly licensed learning resources to the Kolibri offline learning platform. It is designed to be deployed in communities that lack consistent access to the public internet. They currently have a significant amount of English language content, but are limited in what they have available in the dozens of other languages they have translated the core software into. I'm trying to bring in new resources and then try to lead an effort to do translations of the best materials I can find to make the platform more useful to more people.

https://community.learningequality.org/t/bringing-new-comput...

Also trying to recruit people to teach tech newbies how to build their own handheld video game consoles. Let me know if you might like to run a class where you live and i'll share my class materials.

https://community.arduboy.com/t/looking-for-instructors-to-t...

I'm working expanding upon one of the first web app projects I had on my resume when initially starting out in Software Engineering, HoopTime. It's an app for pick-up basketball players looking for a good park for games around your city. Right now focused on cities close to me, namely, Sacramento.

For this iteration of the project, I'm using Manus to build it. My first stab at using AI to build a web application, and the results have been interesting. Although I'm not debugging the code as much with this approach, I was surprised to still feel a similar level of 'fatigue' as I'm guiding the LLM along with the build. Check it out, would love your thoughts!

https://hooptime-vbn5prc4.manus.space/

Building a tool to make test reports useful - https://gaffer.sh

For language geeks: https://kpt.datamediate.com

KPT is a language app specifically targeted at explainable verb conjugation for highly inflected/agglutinative languages. Currently works for Finnish, Ukrainian, Welsh, Turkish and Tamil.

These are really hard languages to learn for most speakers of European languages, particularly English - we're not used to complex verb conjugations, they're hard to memorise and the rules often feel quite arbitrary. Every other conjugation practice app just tells you right/wrong with no explanation, which doesn't really help you learn when there are literally hundreds of rules to get right.

The interesting part was using an LLM to create a complete machine-executable set of conjugation rules, which are optimized for human explainability, and an engine to diagnose which rule is at fault when you get it wrong. There's several hundred rules needed for each language in order to cover all exceptions.

NB as a bonus it also works fully offline because my best practice hours are when I'm travelling and have poor connectivity.

I have been working on quite a bit of things… but lately I went back to my hobby project of the last year:

https://DuoBook.co

It’s like netflix for language, where users can select/create their personal bilangual stories.

I had quite a lot of feedback from HN, friends, random people on the internet and trying to solve the common pain points and find my way around to make it geniunely useful.

- Most people said it’s hard to come up with a story, so I added url grounding. Also added buttons (including HN :)) so people can just click click and get their stories at their level with their interests.

- Made sure people can generate stories without ever signing up

- Each word is highlighted while being read, and the meanings can be checked with a tap. I also added an option for users to read the sentence for being checked how good their pronounciation is.

- Benchmarked 7 different models to get the fastest & highest quality story generation (it’s gemini now) and it’s insanely fast. I might share more about it on the webpage because I am an engineer and I enjoy this stuff lol.

- Added CSV import in Use my words so Anki users can just import their words to study.

- Also people can download their stories as pdf so they can send it to their kindles.

- I am working on a ChatGPT app, so people can just say “@DuoBook give me a Dutch/English story on latest Iranian events” within ChatGPT, but I am a bit afraid that it might be costly lol.