It‘s work in progress but I‘m pretty happy with the outcome so far, especially the data table component.

THINK is a modern CSS-first UI framework built on semantic HTML, custom elements, and data attributes. Uses :has(), container queries, and density scaling. No classes, no build step.

https://think.iotdata.systems/

* Integrating website liveness data into the crawler to make more informed decisions about whether to keep or wipe data from a website if it can't be reached while crawling

* Working out why the liveness data gathering process isn't stopping as scheduled.

* Deploying a voluntary max-charge request header for the commercial API

* Making website URL elements searchable. They should be already, but are not for some reason.

* Maybe looking into an intermittent stacktrace I get on one of the index partitions.

No blockers.

I built a TUI sampler which cherry-picks my favourite features from modern & vintage hardware samplers, DAWs, plugins, outboard FX gear, and DJ equipment.

If you know what an AKAI MPC Live, MPC 3000, SP404, SP1200, BOSS RC-202, Alesis 3630, Serato Sample, S950 filters, and stem separation does, then you'll love seeing these "greatest hits" up in a terminal interface.

Last year while on vacation in Costa Rica, I started scratching my own itch for locating and organizing samples, which quickly evolved into adding more and more features while keeping it tactile and immediate. It was too fun to stop so I kept going. After a few days I was happily making beats in it, and since then it's only gotten better.

It's live and totally free to use, and works in macos & Linux (Windows soon). I'm about to launch v1.0 now, just working with folks in the community to round out the Factory Kits a little more for users new to beatmaking.

Turns out, making beats with no mouse and a terminal interface strikes the perfect balance of hardware feel and software power, and I'm loving the result. Been sharing it with folks in my beatmaking sphere and have plans to continue expanding its reach through more collaborations, contests, and in-person events.

Hope it brings you as much joy as it does to me :)

* https://screenspy.app - observe what youd child is doing on desktop PC. Roblox or homework?

* https://weblock.online - a VERY restricted, whitelist-first mobile browser for kids, use it instead of Safari. I want to feel calm when my children browse the web.

It's a lightweight screen recorder for macOS that lives in your menu bar. It's built with SwiftUI and ScreenCaptureKit, uses the native Content Picker to select what you record, and supports ProRes 422/4444, HEVC, and H.264 — including alpha channel and HDR. Frame rates from 24 to 120fps. System audio and mic simultaneously. You can also exclude specific things from recordings, like the menu bar, dock, or wallpaper.

No tracking, no analytics, no cloud uploads, no account. MIT licensed. Everything stays on your Mac.

https://github.com/Fingel/gelly

I thought it would be pretty simple, but here I am almost 6 months later still adding features. The positive feedback has been nice, though! People seem to appreciate (like I do) that its fast and doesn't use Electron or some other cross platform toolkit. Learning a lot.

It's not vibe coded. Sad that I have to make that qualification these days, but here we are.

The insight: most solo founders need basic "alert me when this changes" monitoring, but existing tools force you through signup flows, credit cards, dashboards you'll never use. So I made it dead simple:

    curl "https://watch.arkforge.fr/api/check?url=https://your-site.com&email=you@email.com"

I built this because I kept forgetting to monitor my own stuff. Now it's live and I'm trying to get my first 5 beta testers. The challenge is marketing - I'm a developer, not a growth hacker. Learning as I go!

What's been your biggest challenge with your current project?

https://github.com/vishnugt/TCPFinMonitor. Live - https://keepalive.gt.ms/

This tool tracks TCP FIN packet timing to see how upstream connections are closing and how keep-alives behave. It helped me spot when connections were closing too early or timing out, which was causing those 503 errors.

https://github.com/vishnugt/hyperbin

A fast, minimal httpbin clone written in Rust. It’s way faster(20x throughput) than the usual httpbin and useful for testing HTTP clients and debugging requests without extra noise.

These aren’t polished, just some stuff I needed to iron out the issue.

I'm building Ditto — it clones websites with 100% visual accuracy and outputs a proper React app with named components and preserved structure.

The problem: you find a design you love, want to use it as a starting point, and your options are either manually recreating it or using a tool that spits out a tangled mess of divs and inline styles. CatchDitto gives you an actual codebase — clean component hierarchy, sensible naming, structure you can extend without wanting to rewrite everything first.

I'm still iterating, would love to hear what others think.

From the README: "[git-forge is a] simple CLI tool for basic interactions with issues and pull requests across GitHub, GitLab, Gitea, and Forgejo".

Right now, I am looking into better testing. Currently, I do testing by mocking the forge APIs and then running git-forge against them with TypeScript. But not everything is testable that way. The TUI is pretty much untested. So I now want to port at least the tests to Rust (I am probably gonna leave the mock API in TS) and need to look into how to tests TUIs, which is a bit of a challenge since not only is Rust my first "systems programming language", I am also not knowledgable in Terminal/TUIs...

It's a bit like Reddit but focused on learning. (Doom learning instead of doom scrolling)

You 1) upload a source 2) direct the kind of questions you want to be asked 3) start answering (and if you get the answers wrong, you can discuss the problem with "AI").

You can read other people's sources, questions, answers and their discussions with AI too.

And if you're learning the same thing as other people, you can join communities to share sources/questions.

It's still very early on, so I'm very interested in any feedback.

I want to create a tool that would automatically block the stealers from stealing your previous credentials or crypto wallets. I had this idea after the Shai-Hulud attack

It's an experimental side project, but so far it looks very promising.

Each game adds more building blocks to the editor: multiplayer, event systems, NPC behaviors, pathfinding, etc. I build a system once, and then anyone using the editor can use it in a click. For game logic, I recently added a visual event system I’m really excited about. It’s kind of like Unreal Blueprints, but focused on 2D. You pick a trigger, wire conditions, and chain actions in a node graph [1].

Big challenge right now: most people who want to make games needs assets, and don't know how to get/make them. So I’m building a marketplace where pixel artists can upload tilesets/characters, and unlike itch.io, assets are usable directly inside the engine. No ZIP downloads or import setup, just browse and drop into your game. A preview here[2].

Also, if you want to use the editor but ship elsewhere, you can export terrain, animations, and hitboxes to Godot 4. Nothing is locked in.

The engine/editor is at https://craftmygame.com if anyone wants to poke around! And you can test a games here[3][4], and 1 multiplayer game I've tested IRL in a bar [4]!

[1] https://youtu.be/8fRzC2czGJc

[2] https://www.youtube.com/watch?v=hScOK_naYnk

[3] https://craftmygame.com/game/e310c6fcd8f4448f9dc67aac/r/play

[4] https://www.youtube.com/shorts/WOIUmOVvaZM

Forge is an orchestration layer that sits between AI coding tools and your codebase. It's a single Rust binary (~3 MB, zero runtime deps) that runs as an MCP server over stdio. Any MCP-compatible AI tool can call it.

MIT licensed. Whitepaper with the full architecture: https://nxtg.ai/insights/forge-whitepaper

A metacritic like website but for any product.

It analyzes thousands of professional critic reviews to find the best of the best.

I started building this because I adore how metacritic analyzes professional movie/game/tv show reviews and calculates a meta score for each title. In my experience it’s the best way to discover new things to watch or play, and I’ve often wished something like this would exist for when I want to buy a product.

This year, I decided to start building it myself and Criticaster is the result.

For a given product category we collect all professional reviews of a given product, analyze each to assign them a score and then calculate an average critic score.

The goal is to become the most trustworthy source to make product decisions.

Very curious what y’all think!

Started with it because I was struggling with finding relevant conversations about my first app where people are exactly asking for what I'm selling, only that I was missing those conversations and people. Build a POC, tested for myself and started getting good leads, so I converted it into my second app.

A PDF generation API, Chrome-based. Most of my time lately goes into print production - browsers render everything in RGB but print needs CMYK with ICC color profiles, and getting that conversion right inside the PDF turned out to be a much deeper problem than expected. Got PDF/X-1a and PDF/X-4 working now.

I got frustrated with existing screenshot services choking on cookie banners, rendering half-loaded pages, and serving bloated images. So my co-founder and I built one that auto-dismisses cookie consent dialogs using Playwright heuristics, serves AVIF-first from Cloudflare R2, and supports geo-distributed rendering so you can capture pages as they'd appear from different regions.

Spring Boot + React + PostgreSQL. Bootstrapped after selling a previous ecommerce SaaS.

Currently documenting the whole build in a 30-day series on the blog if anyone's into that sort of thing.

Supports VS Code, Neovim, IntelliJ/JetBrains Family, Zed, etc.

About to do the first beta release this later this week.

The protocol is "srcuri" (pronounced, "Sorcery")

This site is: https://srcuri.com/

Source code: https://github.com/browserup/sorcery-desktop

https://www.zookeeperapp.com

https://www.mikeayles.com/#zookeeper-wip

It lets you take photos of all the animals you see to collect them, when you 'capture' a new animal, it gives you fun facts about them.

I seeded it with UK zoos, but there's no reason it can't work elsewhere.

It was built because the signage at a zoo we went to was terrible and we had no idea what some animals were, so it matches your photo with the list of animals to the best of its ability.

https://feedbun.com - a browser extension that decodes food labels and recipes on any website for healthy eating, with science-backed research summaries and recommendations.

https://rizz.farm - a lead gen tool for Reddit that focuses on helping instead of selling, to build long-lasting organic traffic.

https://persumi.com - a blogging platform that turns articles into audio, and to showcase your different interests or "personas".

Branch targets are encrypted — decryption only works if the chain state is correct, which requires every prior control flow decision to have been correct, all the way back to the run key.

I'm using single round AES instructions which require AES-NI Wrong key, modified binary, or even a debugger breakpoint causes execution to instantly collapse into garbage. Its written in Rust and currently passes 261/276 Wasm spec tests (remaining 15 are cross-module imports).

Currently supports SIMD, atomics, exception handling, memory64, tail calls, and WASI with sandboxed file I/O and TCP sockets.

Im tinkering around with porting it to a sel4 backend in adition to the posix backend.

https://www.inclusivecolors.com/

Unlike most tools based around autogenerating colors, this is more of an editor that lets you fully customise all the tint/shades to your liking with a focus on accessibility. This is important when you've got existing brand colors to include and want to find accessible color combinations that work together.

Would love feedback in general and especially from designers/devs who have different needs in how they go about creating branded palettes!

The game is about spacepirates playing basketball, it's kinda a basketball manager game. It's played in your terminal and works with no internet.

You can try it via ssh at `ssh frittura.org -p 3788`

- https://github.com/rumca-js/Internet-Places-Database - map of the Internet domains

- https://github.com/rumca-js/Internet-feeds - database of RSS feeds

- https://github.com/rumca-js/yafr - very simple RSS reader

- https://github.com/rumca-js/crawler-buddy - crawling project

- https://github.com/rumca-js/Django-link-archive - another RSS reader

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, curated by you version of the Daily Mixes. You can add some advanced filters as well, if you really want to customise what music you'll get.

It works best if you follow a good amount of artists. Optionally you can get recommendations from artists that belong to playlists you follow or you've created - if you don't follow much or any artists, then you should enable that in order for the service to be useful.

https://riffradar.org/

`brew install tritium` (macOS)

`winget install tritium` (Windows)

`curl -f https://tritium.legal/get | sh` (linux)

Check it for free out and let us know your thoughts!

An opensource iot drone for less than thirty dollars.

It's an infinite canvas that runs SQL.

I've been working with data my entire career. I feel like we need to alt+tab so much. What if we just put it all on a canvas?

Currently very WIP, but there's a simple titanic demo available!

Built with tldraw and duckdb wasm, running on cloudflare durable objects

It's a free macOS app written in Swift that allows you to type with your voice. It supports local models and BYOK (Bring Your Own Key) with a bunch of providers.

You can assign different models and post-processing steps to polish the text. For example, I have a setup for Obsidian that transforms my voice into clean, formatted Markdown. Or, when I use it inside VS Code, it switches to the Parakeet V3 instant local model.

https://www.mikeayles.com/#phaestus-wip

It is capable of creating a PCB (and outputs gerbers, bom, pick and place files), an enclosure (written in SCAD, outputs an STL for printing), and firmware, which it's able to compile using a pio runner on railway and provides a binary, but also has a webserial flasher for ESP32's.

There is a blog here, but i've been focussing on getting things finished, as I built it for a hackathon ending today.

https://phaestus.app/blog

I need to update the blog & writeup, because I have the first product it created, a bluetooth remote control. It wasn't without issues, but I have a working PCB, in an enclosure that was printed from it's design, running firmware it generated.

https://photos.app.goo.gl/4r4XtQ65rx9iBUVp9

I would really like people to try it out and report bugs, failures, and successes.

https://github.com/Vibecodelicious/opencode/blob/surgical_co...

I'm currently trying to get the LLM to be more proactive about removing content that is no longer useful in order to stay ahead of autocompaction and also just to keep the context window small and focused in general.

A high-performance 3D game engine and editor in Rust. It has the ability to deploy to WebAssembly and WebGL2, delivering console-quality visuals and near-native performance right in the browser.

Currently building a multiplayer cozy farming game, inspired by Animal Crossing. Reach out to our discord if you are interested in learning more: https://discord.com/invite/mHsQayQNdp

https://www.fastpause.app/

Offline first, no tracking PWA for intermittent fasting and mindful eating. It helped me lose another 3 kg in January. Spiked a native iOS version, but I really like the simplicity of just the PWA. Not sure what's next!

--

https://reposit.bot/

Having done a lot of back and forth with LLMs and at the end throwing away learnings from a conversation felt so wasteful - reposit allows you to /share a summary of the valuable learning from your LLM chat for others to discover.

At the beginning of researching a problem, your agent can search reposit just like Context7 for docs. This way, even if you opt out of sharing your data with your LLM provider (as it's all or nothing), you can choose to publicise a solution to your problem with very little effort.

I'm working on extracting valuable learnings from open-source community projects as a starting point now (with attribution), as it probably needs a larger database to be valuable for users to install and use.

You can also self-host it and share privately within the company.

I train BJJ and kept hearing the same pain points from academy owners regarding attendance tracking, communications, missing payments, etc.

So I built a tool for martial arts academies in 2024 with belts progression, automated payments, attendance tracking, and a tablet check-in system. Nowadays I'm still onboarding new academies every week and working a bit more on the marketing side to keep growing.

https://grumpypants.bandcamp.com/track/space-amoeba

The idea is you take a book you want to read, and it gets translated but also rewritten to match your current learning level. And as you read/listen it introduces new words to learn, reinforced by spaced repetition.

We're taking a trip to France this summer and I'm hoping to have something usable for at least a couple months before we go.

Currently working on the mechanics of extracting content from ebooks.

https://www.pkgstore.io

NuGet.org doesn't distinguish between a hobby project and a professionally maintained library with real support. pkgstore is a curated directory and marketplace where publishers can sell NuGet packages directly, with full dotnet push/restore support, Stripe payments, and automated access control.

In open beta now, onboarding publishers. Would love feedback.

The first three are:

- miniWake: keeps the computer awake

Alternatives: Powertools; USB mouse jigglers

Features: installs without admin rights; triggers invisible mouse events; turns off at LOCK, turns back on at LOCKOFF (saves battery); manual turn off or on via double-click on the icon

- miniRec: records system audio + microphone to mp3/wav

Alternatives: various utilities like Voicemeter, AudioRouter, or some DAWs

Features: does not require any special driver; installs without admin rights; light on resources; "invisible" to third parties (video meetings); auto turn off after 5 minutes of silence (configurable)

- miniCron: system scheduler as a service

Alternatives: NSSM - the Non-Sucking Service Manager; Splinterware

Features: launches any program at any given time (cron like but without cron syntax); kills the current task when the service is stopped; reads and logs stdin/stderr; very light on ressources and very simple

Two others are in the works.

In the above article, we list a few applications that we think this could be helpful for: life skills, management/sales training, personal coaching, etc. We'd love to demo the software if this sounds interesting to you!

https://github.com/alsoftbv/topic-lab

It's a Tauri-based app so it has small binaries and supports MacOS, Linux, and Windows.

No screenshots yet, couldn't find the time for marketing work. I'm building features as I am using them. I wanted my colleagues to give it a go first before sharing to the public, but I believe it's already valuable as-is.

[0]: https://github.com/zikani03/basi [1]: https://code.zikani.me/using-the-zig-built-lightpanda-browse...

I've been working on my newsletter called Tech Talks Weekly[1] where my readers get one email per week with all the latest Software Engineering conference talks and podcasts[1] published that week.

In January, I've released a paid tier[2] where my subscribers additionally get:

1. Access to my internal database of all the talks and podcasts since 2020 (+48,000 in total) where they can search, filter, sort, and group by title, conference/podcast, view count, date, and duration.

2. The list of the most-watched talks over the last 7, 30, 90 days, 6 months, and 12 months based on number of views.

3. Category-based view of new talks & podcasts by tech stack, language, and domain (Software Architecture, Backend, Frontend, Full Stack, Data, ML, DevOps, Security, Leadership and every major language & ecosystem)

[1] https://www.techtalksweekly.io/p/what-is-tech-talks-weekly

[2] https://plus.techtalksweekly.io/

Currently building chess puzzles based game called ChessBingo[2]. It's almost finished, but there are still things to polish.

[1] - https://onlinefreesolitaire.com

[2] - https://chessbingo.com

Currently in Shanghai but will move to Tokyo next week. Once I'm in Tokyo I'll publish a few posts about AI assisted coding and product creation.

Also adding a few things to my ideas page: https://bryanhogan.com/ideas

Other things I'm working on:

- https://dailyselftrack.com/ - Got into working on it again, mainly solving some UX problems currently.

- https://game.tolearnkorean.com/ - Learn Korean words quickly, words go from easy tasks (e.g.) matchings pairs) to more difficult ones (writting it), currently still needs some slight adjustments, and then I'll release an Android version.

- https://app.tolearnjapanese.com/ - Wanted to learn Hiragana quickly, used my existing project as a base to build this. Needs some adjustments as well, feedback is highly welcome.

- https://tolearnkorean.com/ - Since I'm learning Korean, and also working on an app to better learn Korean, I also want to make a guide on learning Korean, improving my own skills by teaching others.

Some stats so far:

- 200 users

- 378 startup jobs

- 500+ posts

- 2800+ funding rounds

- 1700+ startup companies

- 5000+ investors

The next part of the project is Tech Posts Intel: a lead gen tool using statistical methods to predict which companies should have a funding round coming soon. I'm hoping to soft launch it this week.

I'd love to hear anyone's feedback on the website. Advice on how to get inbound links in 2026 would also be greatly appreciated!

Orange Juice

https://oj-hn.com

[1] https://news.ycombinator.com/item?id=40717398

Basically, if you define a data model with bindings, you can inject data into it or extract data from it by running SAX-style visitors. You can use serializers/deserializers for standard formats like JSON/BSON/CBOR/CSV, or you can define custom formats for formating structured data however you want to. You can also run a serializer visitor on a deserializer to convert between formats. You can compose filter visitors to extract a subtree or filter out keys. And it's designed to fit on microcontrollers with very limited dynamic memory allocations, because it either streams data on-the-fly or works directly with the underlying data format in a big preallocated buffer.

I worked with libraries that offered a subset of these features before in my professional career (even built one myself), but recently I've had an epiphany (a document can also be used as a data model) that makes me think I can create something elegant and unique.

The platform itself is built on elysiajs/bun and tanstack and is completely hosted in EU and the payment processor is a EU based entity and we have an ISV partnership.

Biggest challenges: - How to explain the different use-cases/possibilities in a clear way - DX for any hacker who comes across the device with/without hardware experience

Most recently, we added support for benchmarking (create custom dashboards tracking nav and value chart of subsets of your portfolio) and us stocks, etfs etc.

I wrote about benchmarking here: https://finbodhi.com/docs/blog/benchmark-scenarios

NOTE: you can try demo without signup, but it doesn't work in Firefox Incognito mode.

Right now I’m struggling to beat a baseline LightGBM model trained on hand-engineered expert features. My attempts at using a win probability head on top of nanoGPT, treating events as tokens, have been significantly worse. I am seeing about 65% accuracy compared to the LightGBM’s 70%. That 5% gap is huge given how stochastic the early game is, and the Transformer is easily 4 OOM more expensive to train.

To bridge the gap, I’m moving to a hybrid approach. I’m feeding those expert features back in as additional tokens or auxiliary loss heads, and I am using the LightGBM model as a teacher for knowledge distillation to provide smoother gradients.

The main priority here is personalized post-game feedback. By tracking sharp swings in win probability, or $\Delta WP$, you can automatically generate high or low-light reels right after a match. It helps players see the exact moment a play was either effective or catastrophic.

There is also a clear application for automated content creation. You can use $\Delta WP$ as a heuristic to identify the actual turning points of a match for YouTube summaries without needing to manually scrub through hours of Twitch footage.

It synthesizes comments into structured reports in a Chief-of-Staff style for tech leaders.

https://hnsignals.com/signal/46937696

After one year of development, it's going better than I expected, so I'm considering building a demo to gather feedback and see if there's enough traction for working towards a Steam release.

Even if that's not the case though, it's been a blast learning about game dev in Unity/C#, as well as 3D modeling and animation in Blender!

I'll admit it's terrifying to share this here because I don't know how to keep costs under control. For now only myself and my friends have used it.

Also trying to make a podcast out of it, which I enjoy listening to while I do some road trips: https://open.spotify.com/show/1fFwWMWJBJYIZmyz9cnrKB

The site has become quite a hit and gets thousands of unique visitors each day. https://www.privateequityvet.org/vet-list

Inspired by a TED talk I saw [0] where the researcher from Microsoft displayed a program with AI assisting with thinking while someone was reading and annotating a document. They claimed it was a way to sharpen critical thinking instead of killing it. They didn't release the product, but I figured it was cool and useful, so I've spent the weekend creating it. It's been a great way for me to practice using agents, and I've learned a lot from this process.

[0]: https://www.youtube.com/watch?v=3lPnN8omdPA

We need to reduce the entry barrier (it's meant for companies so it needs explicit registration) so anyone can use it as a proper SaaS but so far we already have a couple clients :D

That means I have to: - build something so I can evaluate the results. - track each of these projects separately otherwise they turn into dust after quite some time. Gladly claudesidian seems to be working well with the unstructured stream of inputs. Feel like hooking it up with some task tracker cli and calendar and notifications could make life a bit better too. - plan next projects to keep evaluating other skills and tools

It’s been discussed so many times the amount of new or personalized software that appears and will appear and it seems so true.

Whatever I built I am actively using myself - a text rewriter that cleans some of the AI speak and has MCP and cli (at https://www.refineo.app). Math teaching and solving extension at https://math.photos and a self hosted stock opportunity discovery tool that runs locally. This is just to automate what I did before manually and scale it up a bit.

> Any new ideas

There’s no product yet to cover the needs of all of us launching the software into the internet void. Any ad platform out there is a hot and very outdated mess and I just can’t. There is going to be a better way with all the capabilities we have and someone is going to really nail it.

Currently it has:

- Accurate recreation of Windows shell with start menu, taskbar, windowing system.

- Full desktop themes customization (color, cursor, sound, wallpaper, screensaver). All Win 98 default Plus! themes are included.

- Persistent local file system & mounting local folder as removable disk with ZenFS.

- Support playing Flash games and run DOS games (save game persisted). Yes, you can play Doom and copy your savegames to continue.

- Some accurate remakes of Windows 98 apps, some made by me (Solitaire games, Minesweeper, Notepad) some are existing ports (Pinball, JSPaint, Webamp, etc).

- Some other fun stuff

If you're interested in Windows 98, this is for you. You're also welcome to contribute or fork it to create your own version: https://github.com/azayrahmad/win98-web

More about the product itself: https://kraa.io/about

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

https://skicompromise.ch

This was a fun little project I did over the Christmas holidays but only finished off recently. Basically I precalculated the public transport time between the most populated towns in Switzerland to every ski field (about 350 of them!) and then built a little web app around it using Django.

You can choose to prioritise shortest (lowest time overall) versus fairest (smallest variance in group members).

Totally free to use. Next steps are to integrate it with live snow conditions/open lifts...

Claude did help a lot with the FE part. The biggest part was actually finding the best public transport stop for each ski field - that was a very manual process trawling through skimap.org images and Anreise info on ski resort websites.

https://metricme.app/

Over two decades ago I was diagnosed with high blood pressure (for which am I have been on meds for about 15 years). I also have low platelets (red blood cells, basically means that I bruise easily and that small cuts don't heal fast). At any rate, I do blood tests on a regular basis to keep things in check. I have been keeping track of test results, weight and blood pressure result for nearly 20 years, but the data lives in a text file on my desktop. I wanted to build something more substantial for this for quite some time now, so, this is it.

https://play.google.com/store/apps/details?id=com.superfun.w...

https://memory.store

The problem: if you use multiple AI tools (Claude, ChatGPT, Cursor, etc.), none of them know what the others know. You end up maintaining .md files, pasting context between chats, and re-explaining your project every time you start a new conversation. Power users spend more time briefing their agents than doing actual work.

Memory Store is an MCP server that ingests context from your workplace tools (Slack, email, calendar) and makes it available to any MCP-compatible agent. Make a decision in one tool, the others know. Project status changes, every agent is up to date.

We ran 35 in-depth user interviews and surveyed 90 people before writing a line of product code — 95% had already built workarounds for this problem (custom GPTs, claude.md templates, copy-paste workflows). The pain is real and people are already investing effort to solve it badly.

Early users are telling us things like one founder who tracked investor conversations through Memory Store and estimated talking to 4-5x more people because his agents could draft contextual replies without manual briefing. It helped close his round.

Live in beta now. Would love feedback from anyone who's felt this pain! :)

https://store.steampowered.com/app/4009620/Tutankhamun__Buil...

It's inspired by a moment a few years ago when I realized I had no history of what I had worked on in the past.

It let's you quickly get answers to questions like:

- What did I work on last week? - What was that one hacker new post about compiler optimization that I forgot to bookmark last week?

And it has MCP support so it plays well with Claude.

I've used it recently to target specific job applications by adjusting my resume based on what the job application is looking for and what I've worked on in the past... Claude one shots this (because it has context from Memex). And it feels amazing!

Also, the name Memex comes from Vannevar Bush's 1945 article "As We May Think". He originally thought of as a device in which individuals would compress and store all of their books, records, and communications, "mechanized so that it may be consulted with exceeding speed and flexibility". The individual was supposed to use the memex as an automatic personal filing system, making the memex "an enlarged intimate supplement to his memory".

And he created a word - memex - which is a portmanteau of "memory" and "index".

The wikipeida entry here has more information: https://en.wikipedia.org/wiki/Memex

It's been a slow start for me. But now, between the cli interface and the MCP connection with Claude I find that I'm starting to use it instead of:

- Bookmarks - Lists / Bug Tracking

And even more exciting it's unlocking capabilities that I didn't have before:

- Can ask Claude to review the last week of work and remind me of things that I might still need to do - Can prevent randomization when someone asks me how to configure a server that I set up a month ago. Now I just ask claude and it checks in Memex. And I can send over a nice .md file

An AI based time tracker: reconstructs your day from whatever it sees you doing. Screenshot based but never stores them.

https://donethat.ai/data

The same tech stack is pretty easily adaptable to openclaw tracking. If anybody would like to try, DM

Also looking into AI based security tools for monitoring security of DoneThat. Thinking of using zeropath would love to hear if people tried them / have other suggestions

It does this by taking a screenshot of the page, but before it does that, you can modify what’s displayed in the screenshot with CSS, tailwind classes, meta tags, or HTML templates.

If you connect your website to it, the only thing you need to deploy to your web app are a few meta tags. The OG+ servers do the heavy lifting of processing the meta tags to setup the page, take a screenshot of it, and serve it up to the consumer.

The other cool thing it does is generate a different Open Graph images per social network so they all get an image for the exact size they works best in their previews. The CSS or HTML templates are aware of this too so you can display different content to specific social networks.

How Home Alone My House - A fun app I'm making with my children using computer vision. The idea is I can scan the room with my camera before unwittingly walking into their traps and becoming a hapless adult who didn't pay close enough attention to tripping hazards and choke lines.

https://github.com/asteroid-belt/skulto

Started building this after getting nervous about installing random SKILL.md files from GitHub. Scans for prompt injection in markdown/references and suspicious patterns in scripts/.

- 200+ curated skills included

- 33 supported agents

- Symlinks for one install anywhere and automatic updates

- CLI, TUI, or MCP interface: try asking Claude to find and add Awesome repos.

- Semantic search across skill content

Working on: local skill authoring, mise-style directory activation

Go + Bubble Tea. Happy to hear what's missing.

Website: https://hobbyboard.aravindh.net

GitHub: https://github.com/aravindhsampath/hobbyboard

I want to do a show HN later this week.. but here might be a softer launchpad :-)

The second bubble there is a tool for 3D visualization and analytics of Claude Code sessions. The sample conversation is the one that made the tool itself!

That was a fun toy I learned a lot from. I’m not expanding that but am working intensely on the first bubble:

thinkt a CLI/TUI/Webapp for exploring your LLM conversations. Makes it easy to see all your local projects, view them, and export them. It has an embedded OpenAPI server and MCP server.

So you can open Kimi and say “use thinkt mcp to look at my last Claude session in this project, look at the thinking at the end and report on the issues we were facing”.

I added Claude Teams support by launching a Team and having that team look at its own traces and the changing ~/.Claude folder. Similar for Gemini CLI and Copilot (which still need work).

Doing it in the open. Only 2 weeks old - usable, but early. I’m only posting as it’s what I’m working on. Still working on polish and deeper review (it is vibe-crafted). There’s ergonomic issues with ports and DuckDB. Coming up next is VSCode extension and an exporter/collecter for remote agents.

https://github.com/yelinaung/expense-bot/

https://gitlab.com/yelinaung/expense-bot/

As you may see from the git history and "contributors", it's mostly Claude and AMP making the changes. I am not entirely sold on these agents and not particularly excited by these. But I also feel that I can't afford to sit out this transition so here I am...

I'm a self-taught coder who first built this 7(!) years ago but couldn't figure out the OCR part. Started again 9 months ago on Replit (starting with Agent 2 which was okay, then eventually starting to absolutely crank with Agent 3) and it works really well now.

Would love feedback from any golfers! golfrise.com

Hosted OpenClaw, one click and you get a full agent with configurable skills, channels and the whole thing, all running in its own sandbox.

I love OpenClaw but setting it up is a pain: VPS, Docker, API keys in plaintext, security patches... So I’ve spent the last couple weeks building a hosted version that handles all of that. Each user gets their own isolated environment on Cloudflare Workers.

Still doing some testing with friends before opening signups but planning to launch properly this week.

Would love feedback on the landing page in the meantime!

Some technical highlights:

- Graph-based signal flow: Products become nodes, connections are edges inferred from port compatibility (digital, analog, phono, speaker-level domains)

- Port profile system: Standardized port definitions (direction, domain, connector, channel mode) enable automatic connection inference

- Rule engine: Pluggable rules check completeness, power matching, phono stage requirements, DAC needs, and more

It's getting close. I'll do a show HN on it sometime soon.

https://designflo.ai

This is a follow up to an idea I had years ago https://news.ycombinator.com/item?id=13022649, which is now semi-automated (with lots of manual curation as the last step).

The biggest challenges:

- how to organize all this info in a nice way

- where to find more time to read all the gems I've found so far :)

UPD: formatting

Aiming to get that published in the next day or two, and then I plan on diving in on a complete rewrite of the book I wrote on building Shopify apps with .NET and C#. It's long overdue, the book still uses Shopify's deprecated rest API and some methods that aren't supported anymore, but I've been holding off on an update until I could rewrite it with the new fluent query builders in ShopifySharp.

Outside of my OSS stuff, I'm continuing working on my SaaS app, Stages (https://getstages.com) [¹], which has been paying my mortgage and bills. Customers have been asking for lots of features lately and I'm anxious to get a particular one finished (filtering orders and events before they come in and are saved to the app) soon. It's my biggest source of churn right now.

[¹] Elevator pitch: the app is like a pizza tracker for your orders that have a custom or long, drawn out production process. Your staff and customers can see exactly where an order is in the process without calling or emailing you. Shopify only for now but one of main dev goals is to move beyond Shopify.

Behind the scenes I’m rebuilding the sync engine to properly support offline mode. Trying to get to instant opens for the app (and of course work offline). It’s probably my 5th sync engine. It’s been really fun to see how much easier, faster, better, etc each new iteration is.

(And the project at large is https://phrasing.app - a language learning app for polyglots. It’s like anki but designed to be enjoyed)

The biggest problem is internal knowledge and external knowledge systems are completely different. One reason internal knowledge is different it is very specific business context and/or it's value prop for the business that allows charging clients for access.

To bridge this gap, the best approach is to train agents to your use case. Agents need to be students -> interns -> supervised -> independent before they can be useeful for your business.

https://github.com/rush86999/atom . it's still in alpha.

RClone is doing the heavy lifting (amazing project). I'm wrapping it with the operational features clients have asked me for over the years:

  - Team workspaces with role-based access control
  - Notifications – alerts on transfer failure or resource changes via Slack, Teams, Discord, etc.
  - Centralized log storage
  - Vault integrations – connect 1Password, Doppler, or Infisical for zero-knowledge credential handling
  - 10 Gbps connected infrastructure (Pro tier) for large transfers

My most fun feature: when I connect the app to my car, I can use the skip buttons on my steering wheel to rewind or forward 10s in the playback.

A simple one click just works mic+PC audio recorder for Windows that mixes the microphone into the PC sound.

Asterbot is a modular AI agent where every capability (such as tools, memory, LLM provider etc.) is a swappable WASM component.

Components are written in any language (Rust, Go, Python, JS), sandboxed via WASI, and pulled from the open asterai registry. Think microkernel architecture for AI agents.

Month 2 of building the SQL client I've always wished I had.

One feature I'm especially proud of is the visual query builder. Drag & drop to build SQL queries.

There's also an entire SQL tutorial section for anyone who wants to learn or refresh SQL knowledge.

I used to be a coding bootcamp instructor, TA and guest lecturer. I've noticed more and more people need to learn SQL for various different reasons. I'm mostly concerned about lesson scaffolding since most SQL courses don't do it that well. I'm hyped about AI but they're not great with lesson scaffolding.

I'm 33% to 50% done. I've already noticed the way I scaffold the lessons is unconventional. For example, for the first 50%, I don't want students to know what tables are. It's too much all at once, everything should be small bites before the big concepts get introduced.

If anyone is interested in testing the beta version, let me know. It will be up within the next 2 weeks probably. My email is in my profile.

Right now, I'm working on the OpenClaw-like feature. So, you can learn Japanese via Telegram. Keep track your progress. Practice conversation with your AI assistant. Etc.

Game idea: DroneCraft is a third-person drone exploration game where players scout the world for parts, craft powerful upgrades, and trade strategically to evolve their build.

Whats coming: Core mechanics are up and running. First playable version planned within a month, alongside open-sourcing the full codebase.

My main goal is not just a "the model made code, yay!" setup, but verifiable outputs that can show degradation as percentages.

i.e. have the model make something like a connect 4 engine, and then run it through a lot of tests to see how "valid" it's solution is. Then score that solution as NN/100% accurate. Then do many runs of the same test at a fixed interval.

I have ~10 tests like this so far, working on more.

It's a (now more than 20 years old) TV tracking website and community.

I've been using Claude 4.5 Opus (now 4.6) more and more these days modernizing and redesigning sections that haven't been touched for a decade or two. I don't trust LLMs much, but by breaking the work into small, self-contained tasks and testing constantly - I'm making surprisingly fast progress.

Next month prep starts for finding dev work after an extended hiatus.

A developer platform for AI image generation that includes observability, with fine-tuned vision models as a judge to monitor production traffic. (Still working on the last part.)

We also have a model arena and showdown page that ranks models by task, so you can find the best model for e.g. making infographics: https://lumenfall.ai/leaderboard

We just launched the MVP. Tech stack is Rails for the dashboard and Cloudflare Workers (Typescript / Hono) for the gateway.

Since the ShowHN thread, I received more than 40 individual game submissions!

To give more exposure to some of the games launched during the week I also launched a newsletter. Feel free to check it out if you want to learn more about games shown over the week :)

https://hnarcade.com/newsletter

I've been working on an 'anti-fantasy' football game where you pick matches instead of individual players.

You can create or join leagues to compete against your friends.

I’d love to get some feedback!

2. A “runtime scheduler for humans” that I wished existed, too (think morning routines, travel checklists, and pomodoros in the same abstraction—but also a lot of support for ad-hoc rearrangement and addition of the task queue).

    A performance comparison of four common Go string building methods.

I recently updated my go-stats-calculator to include many more stats [2]:

    CLI tool for computing statistics (mean, median, variance, std-dev, skewness, etc.) from files or standard input.

I also created claude-image-renamer [3]:

    AI-powered image renaming script that generates descriptive filenames for screenshots. 

[1] https://github.com/jftuga/go-string-concat-benchmark

[2] https://github.com/jftuga/go-stats-calculator

[3] https://github.com/jftuga/claude-image-renamer

I used this as a real end-to-end project to sharpen my backend skills in Go (API design, data modeling, deployment), while also experimenting with AI-assisted development. It’s live, and I’ve already made a few organic connections through it.

Basically tracking where my friends and I have collectively been by dividing the global map into H3 hexagons. The using photo and workout metadata to get the locations, giving points and doing comparisons between everyone. It’s actually quite fun to see random people around the world sign up and see in the global map where everyone has been. Grounds me a bit haha.

It's a modern, powerful, and user-friendly web interface for managing and monitoring ClickHouse databases. It provides a comprehensive suite of tools for developers, analysts, and administrators to interact with their ClickHouse clusters efficiently.

- Effective and Native RBAC: Use ClickHouse grants to control the data access and UI permissions. - Discover - Flexible, Kibana-like data exploration for any table granted access. - SQL Console, Monitoring and Logging, ... all in just one place.

https://github.com/jgbrwn/vibebin

It automates installing and managing Incus, Caddy, and SSHPiper, provides a TUI for container lifecycle and quick actions, a web admin (built/compiled on the container) for toggling and updating AI coding tools (Shelley, Claude Code, OpenCode, etc.), and a background sync daemon that keeps Caddy routes and container metadata in sync.

Each container exposes coding tool web UIs on isolated ports and supports direct SSH/VS Code Remote access, so you can run multiple independent coding agents against real project files without exposing your local machine.

The project emphasizes simplicity and recoverability for running agents locally: containers are persistent, optionally routed via reverse proxy with basic auth, and tracked in an SQLite DB so setups auto-heal after restarts. It’s written in Go, includes an install script for one-line deployment, targets modest VPS specs (4–8GB RAM recommended), and bundles helpers for DNS and provider automation.

Ideal if you want a lightweight, opinionated way to host multiple isolated AI dev environments on your own server instead of relying on hosted agent platforms.

traymd: A system tray notes application that supports basic live input of markdown. https://github.com/rabfulton/TrayMD

reelvault: A local film browser and launcher. https://github.com/rabfulton/ReelVault

preditor: A simple image viewer that shows each image in the center of the screen in a window sized for that image with some basic editing functions built in. https://github.com/rabfulton/preditor

https://www.pencild.co

https://shovel.js.org/blog/introducing-shovel/

https://github.com/bikeshaving/shovel

> Dev Cleaner is a desktop application for scanning and cleaning development cache files and build artifacts. It helps developers reclaim disk space by identifying and safely removing caches like node_modules, .cargo/registry, .npm, and other build artifacts.

It's closed source, as I am planning to sell a license. But if you email me, I am happy to provide a build.

Klondike solitaire game using Godot. The goal is to better understand Godot's inner workings, and not using any LLMs... outside of whatever Google searches automatically popup when I have questions.

Secondarily, decompiling the DuckTails Gameboy ROM with PHP... then seeing about using PHP to create a GameBoy game. For no reason than to see if it can be done.

https://vimgolf.ai

To show newbies how to use vim. Currently its not complete and has major issues. So if you want to try give it a go, but please hold your judgement as not all shortcuts have been added.

Any issues opened on the repo are most welcome.

https://storystarling.com

Working on a platform where you describe a story concept and it becomes a real, illustrated picture book - professionally printed and shipped to your door.

The key difference from "personalized" book companies: this isn't template stories with a name swapped in. You bring an idea - maybe a book about a kid with a cochlear implant going to their first day of school, or a bilingual German-Turkish story about visiting grandma's village - and it generates a complete original narrative with consistent illustrations throughout.

You can upload reference photos so characters actually look like your child. Supports 30+ languages including bilingual editions on the same page.

Currently refining the showcase features and adding RTL language support.

  10 MCP servers as device drivers (exchange APIs, browser automation, Apple docs, issue tracking).
  200+ skills as prose runbooks that compose system calls. Agent-mail for IPC between parallel
  agents. A drift detector called "wobble" that scores skill stability using bias/variance analysis.

A platform to efficiently work with any data right in the browser. Like interactively visualizing millions or rows, and at the same time augmenting the data with domain-specific capabilities. For instance, the cheminformatics plugin automatically recognizes molecules and provides proper rendering, substructure search etc. Sort of a Swiss Army knife for scientific data.

Not really a new idea, been working on it for many years already :)

https://subsmith.app/

Fully off-grid using solar, batteries and Starlink for uplink. Focusing on AI inference at the beginning. Currently building our first prototype and testing cooling solutions.

https://solarcube.com

An open-source multi-cloud governance framework powered by a YAML rule engine. We just reached a milestone by adding Database (DB) support.

The goal is to allow developers to audit configuration and compliance directly within DB instances, alongside standard cloud resources (AWS, GCP, Azure, K8s). We’re focusing on keeping the YAML rules as agnostic as possible so the same logic can apply across different environments without rewriting everything. rtk (https://github.com/rtk-ai/rtk)

This is a "scratching my own itch" project born from using Claude-code. While vibe coding, I got frustrated watching the agent spam ls -al or cat repeatedly just to "orient" itself.

It creates two main issues:

    The Token Tax: It burns through tokens for info the agent already has.

    Context Pollution: The context window fills up with redundant noise.

The other thing I built but am less interested in personally just through should exist, is something like MoltBook but for more formal topics like the sciences. -> https://ideas.gd/

I worked with manufacturing companies, and the amount of manual document extraction and manipulation, particularly from accounting documents, was always a large burden.

The goal is upload a document → extract structured fields via LLM → generate new documents from templates. Has a dashboard, with an API, along with a webhook, very much a WIP.

https://fetchtext.io

I made a Wordle-like daily puzzle. Every day a new category matching puzzle comes up for you to solve!

It uses a simple points model instead of streaks or financial-style tracking to make expectations visible, progress clear, and follow-through easier.

In real use it’s solving three main problems: - As a Family Chore Chart — a digital chore system that actually gets kids engaged with responsibilities using points and rewards. - As a Personal Habit Tracker — a way for individuals to organize routines, add notes, and earn points toward meaningful self-defined rewards. - As Complete Homeschool Management — tracking assignments, logging progress by subject, and generating reports and transcripts for multiple students.

It’s entirely web-based (no app download) and works on phones, tablets, and desktops. I’m actively iterating on it based on real use, and it’s been most useful in situations where simpler systems actually get used instead of abandoned.

Check it out https://www.pointwisesystem.com/ Pre-Launch offering 6 months free

GitHub:- https://github.com/adityaprasad-sudo/Explore-Singapore Live demo :- https://adityaprasad-sudo.github.io/Explore-Singapore/

I'm also working on a new strength gains-tracking app that is a lot more intuitive, motivating and friend first. I've been using it with some friends for the last 10 weeks and everyone making is consistent gains. It is my first full PWA, vanillaJs, backend is Lucee & MySQL. Works great on iOS and Android, no one has any complaints. The web stack has come a long way I am probably not going to do a native mobile app for a while. I'll probably make it public in a couple weeks.

I am using KVM from Cloudcone (their virtualization software was hacked about a week ago) and I am using RPI4.

Then I need to set up my old website again, which is a pain in the butt. I hard-coded cron and a git-based auto-deployment feature (I think).

https://memelang.net/ https://arxiv.org/abs/2512.17967 https://github.com/memelang-net/memesql10/blob/main/memelang...

If you’re in sales, a business executive or simply curious about what’s going on around your own startup give it a go.

So I'm building Taskplan (https://taskplan.run) - it's like Ansible, but for people. Build a plan, assign tasks to people or teams, and get a real-time dashboard to track progress as the work happens.

I'd love feedback from anyone who deals with the same issues or works on ops-heavy projects.

Started working on a site to document anti patterns in online discourse. Not quite logical fallacies but more so unproductive expressions that aren’t conducive to pleasant, productive, and focused discussion. The site is a bit rough right now and a work in progress.

I want the internet to be a better place for discourse and I think a reference or guide on anti patterns in replies could help make a dent in the right direction.

There have been lots of cool technical challenges through the whole process of building this, and a very nice variety of different kinds of work.

I'm working towards using the outputs from this language to build out levels and assets for a browser-based game I've been dabbling with over the past few years.

- Life’s Articles, a personal Wikipedia

- Counting Worms, a very fast calorie tracker

- BookTalk, a audio based reading companion for capturing annotations

- Kindle Blocker, a Chrome Extension that earns you minutes on websites by reading with the Kindle app

[1] https://www.bramadams.dev/working-software/

https://github.com/bajor/nvim-raccoon/blob/main/README.md

At the moment working on the 3rd party development tools so in the future anyone can make their game dev dreams a reality and make a simple and fun multiplayer party game for the Gaming Couch platform, ideally in only one weekend!

If you're an interested game dev that would like to beta test the dev tools, hit me up either here, via Discord (link available from https://gamingcouch.com) or by emailing me at gc[dot]community[at]gamingcouch[dot]com!

The TL;DR of Gaming Couch:

- Currently in free Early Access with 18 competitive mini-games.

- Players use their mobile phones as controllers (you can use game pads as well!)

- Everything is completely web-based, no downloads or installs are necessary to play

- All games support up to 8 players at a time and are action based, with quick ~one minute rounds to keep a good pace. This means there are no language based trivia or asynchronous games!

Some of the stuff built so far:

https://github.com/system32-ai/chaos-agents

Working on couple more agents around the same problem statement. It has been fun building it so far.

Myself, mostly. Trying to wrestle with realizing how much time I've not been spending on my supposedly main project[1] and questioning whether it's really worth doing.

> Any new ideas that you're thinking about?

Way too many. Writing todo lists is part of working on myself.

[1]: PAPER, a pure-Python ~(pip/pipx replacement), from scratch with an emphasis on simplicity and elegance. https://github.com/zahlman/paper . There's more locally that I haven't pushed, including factoring some stuff out into a separate project and planning more of the same. But yeah.

Side Note : These posts on HN motivated me to start working on this project. Cheers! to the community.

I've been using Claude Code to spin up apps quickly, and I kept needing the same infrastructure every time - user auth, permissions, usage tracking, job queues. So I pulled it all into one SQL package that lives in Postgres. Now when I start a new app I just tell Claude to use Postkit and all that stuff is already there, no external services to set up. I can focus on the actual product and iterate fast.

It was also a good excuse to actually use stuff I'd studied for system design interviews - Zanzibar-style ReBAC for permissions, a double-entry ledger for usage metering, transactional job queues with SKIP LOCKED. ~15k lines of SQL across five modules, with a Python SDK. The SQL works from any language though.

https://github.com/varunchopra/postkit

https://github.com/jbonatakis/blackbird

At a high level it's my take on how the execution aspect of spec-driven development should be handled. Where as most tools that are popular right now break a spec down into a task list and instruct your agent to work through it in a single session, I am treating agents as stateless. By this I mean a separate (headless) session is started with selected context for each task. This avoids context exhaustion, compaction (and the resulting confusion that can occur), and means that Blackbird can work through effectively an arbitrarily large task list.

Right now it's BYO-spec, but then it:

* breaks the spec down into a dependent-aware plan (DAG) composed of parent and child tasks

* executes tasks one at a time based on their status (ready to execute if all dependencies are marked as completed)

* allows you to (optionally) pause execution after each task to review, approve and continue, approve and quit, or reject the changes altogether

* (soon) treats parent tasks as an automated reviewer for all child tasks and optionally auto-resume those sessions to address the feedback

* and more

It's entirely bootstrapped, and so far I'm quite pleased with it. I also wrote a post[1] today about some of the concepts I had in mind as I was defining the architecture.

[1] https://jack.bonatak.is/blah/killer-context/

- I see a lot error propagation with CUAs

- A GUI is very flakey and it produces a lot action latency

- There're hidden states behind each screen that CUAs simply can't capture

- Token consumption is absurd (but I guess this will alleviate as LLMs get cheaper)

What do you guys think? Any good ideas what'd be a good counter to this?

UK only for now, and very much a “solves my problem” side project, but easily scalable to other countries of the need is there!

https://meat.offer-spider.com

Still iterating on it, including a potential improvement to the (very simple) design.

The idea is that future discovery isn't limited by watch history and users on the platform can curate, showcase and amplify their favorite videos. It is an equal opportunity stage where users contribute to build a time capsule of videos.

If that sounds interesting to you, check it out at http://jadestage.com/ !

Speaking of projects, I’m roughing out a method of pulling cost data for common services (compute, storage, databases, etc) across the three major cloud providers and making recommendations as to where to put things for optimal cost; a key component of a “universal cloud” idea I’ve been kicking around since 2020 or so, where the base cloud services are abstracted away into commodities rather than bespoke products or locked-in vendors. The goal is to basically have something like Terraform that will transpose its code to the destination cloud chosen by the cost analyzer at execution, and eventually auto-migrate load as prices or needs change (e.g., a client churning early and shifting that reserved instance to another customer for a higher margin).

Write once, and trust the pricing model to deploy it where it makes the most fiscal sense. No more learning Azure/GCP/AWS for bog-standard workloads anymore.

https://github.com/rdavison/DXX-Raytracer-ar/releases/tag/ar...

Because they're relatively low-effort (Amazon is terrible for sellers in many ways but man do they provide an incredible amount of infrastructure), that leaves me plenty of time to play with AI, and it just so happens that the business serves as a giant, practical eval as new models come out.

I've been vibe coding apps for internal use and using Nano Banana for listing images and whitebox photos, and more recently I've started to lean on Claude Code heavily as an assistant. It's got API creds for my Amazon account, so I use it for everything from figuring out when I need to reorder to filling out spreadsheets for companies that safety test my product.

And of course I am writing a Substack that I must shamelessly self promote that goes into the practical use cases of AI in my business: https://theautomatedoperator.substack.com/

https://github.com/vladkuz/TenantSaas

https://clodhost.com

Oh, I also used the tech to set up claudecrowd.clodhost.com .. a vps running claude code where anybody from the internet can submit the next prompt!!

One of the projects that features in these notes is the attempt to build a programming language using AI. https://github.com/xvandervort/graphoid

Since I left my last job, I do a lot of writing. I also have a couple substacks. One is a humorous weekly look at science and tech (https://technoscreed.substack.com/ ) and the other is a monthly exploration of history (https://historyroad.substack.com/)

We are building a crowdfunding page for agent-run startups. People can co-create business ideas with AI and vote for the ideas they like the most. Agents then run market research and will eventually prototype the proposed ideas. In the future, we also want people to be able to own part of the agent-businesses they have sponsored.

[1]: https://www.agenticcodingweekly.com/

Launching this tomorrow:

https://www.bettermarkets.app/waitlist

Dimensionally accurate AI 3D modelling. My grandpa has a 3D printer but struggles to use any complex tools. So I am working on this chat interface to allow him to do some simple models.

So far he has triggered more than 150 generations. It’s getting better every model cycle and gives me something I enjoy working on.

Pasture takes each signup, enriches it (title, company size, funding, tech stack, and more), and scores it 0-100 against your ICP. Alerts go to Slack with full context. You can also track which channels bring quality vs. junk over time, which has been the most useful part so far.

Built it because most survey tools felt overgrown for what I needed. It focuses on post-purchase and on-site surveys, attribution questions, and getting clean data out.

Lately I’ve been working on:

Simpler targeting + survey logic Exposing survey data to AI tools Improving response rates without nagging users

It’s bootstrapped, profitable, and built by one person (me).

https://kiln.bot

Uses your local Claude Code as the agent and GitHub as its UI, things you already have. Open source, MIT License.

You move cards across kanban columns (Backlog -> Research -> Plan -> Implement) and Kiln runs Claude locally, opens PRs, and keeps everything tracked in GitHub.

I need it to create Gamedev and 3D artists oriented tool for creating SDF-based shader visualizations (with 3dgs/nerf compilers)

90% is done

https://github.com/pavel-voronin/sliced-areas

https://github.com/imbue-ai/ratchets

WIP language spec: https://gist.github.com/Heathcorp/13fcd206fdc38ca6ce001f32ef...

Writing the compiler/solver in Rust with no AI assistance because this is a learning project.

So of course no journal or conference is in the least bit interested, and I'm now reformatting it for another obscure low-tier journal that no-one will ever read.

Otherwise:

- automating the translation of a Byzantine Greek work that has never been translated into English before. https://stephanos.symmachus.org

- also preparing evidence for a case against the university I sometimes work for.

[+] Linear regression, but instead of minimising the Euclidean distance, minimise the p-adic distance - https://en.wikipedia.org/wiki/P-adic_valuation

A Windows 95-themed interactive guide on agentic AI coding, with a hidden SkiFree game, original chiptune soundtrack, achievement badges, and a Red Pill / Blue Pill choice that can BSOD your browser. Seven chapters with a codebase readiness scorer, ROI calculator, and copyable artifacts for engineering leads.

Built entirely with Claude Code, which is fitting since the guide teaches the same workflow. It's a labor of love that happens to be made with the tool it's about.

https://euzoia.substack.com

Full project: https://euzoia.org

Tried to be super low-tech: Notion, super.so, Spotify creators, riverside.

Now thinking of building an email-based agent for behaviour change accountability. Would love any pointers to good UX for email-based AI assistants.

Kinda like HN meets Pocket.

It includes a Chrome extension to easily tag, save & share pages.

Currently the front page is all the pages I find interesting (AI/Startup related).

Would love any feedback or feature requests!

and also Backseat Writer, a creative writing text editor that uses AI to impersonate your audience and give you feedback https://backseat-writer.vercel.app/demo which is more of an anchor for my own writing practice than anything else, but I find it fun

Already have my own JS engine & the basics of three.js and pixi.js 8 working, roadmap to v1.0.0 posted in github issues. Aiming to show it to folks at GDC in March.

Also, watching a bunch of videos and reading docs on OpenClaw. I had thought I'd do an install of it sometime this weekend, but I don't know if I'll get to that at this point or not.

And lastly, messing with Spring AI[2]. I wanted to get a local build of that going so I can dig into the bowels of it and hack on it a bit. So I got that repo cloned and ran a quick build, and now I plan to start exploring the codebase.

[1]: https://github.com/mindcrime/A2ASandbox

[2]: https://github.com/spring-projects/spring-ai

Initial results are promosing Extracting the text and figuring out which lines belong to the same paragraph and then try to map those to the original positions in the PDF...

It generates dashboards automatically, you just point it to your data. It also has a visual editor to adjust layouts, charts, and other dashboard elements.

Gata is an open source automated L1 ticket triage tool for Zendesk. It costs pennies per ticket for it to route tickets to the correct team.

During development I was regularly seeing over 90% accuracy. The average for humans is 60-80%.

The whole thing runs in your AWS account.

There's more information in the release announcement - https://www.proactiveops.io/archive/meet-gata-the-automated-...

Trying to be much more though. Creates an abstraction over all the music streaming services so you can share playlists with anyone, regardless of what subscription they have.

https://greenmtnboy.github.io/space_reporting/

I'm an physician who previously had wrist tendinosis and carpal tunnel and made the keyboard for myself. I'm trying to get the keyboard registered as a medical device for treatment of hand/wrist repetitive strain injury. Currently getting design for manufacturing finalized, and waiting on injection mold prototypes. Hoping to launch on Kickstarter in the next few months.

Also concurrently waiting on ethics approval for a clinical study, which will happen after launch. We had quite promising results from user testing, so I'm cautiously optimistic about the study.

EDOS was a direct 6800 port of FDOS. FDOS was the first DOS available for microcontrollers, using iCOM's FD360 8-inch floppy drives.

https://github.com/jhallen/exorsim

https://www.youtube.com/watch?v=TpHKygZ7OHY

It's about diabetes management. website is done wirh kirby cms.

https://workglow.dev/

https://epilog.tv/

The UI/UX is a pretty interesting problem. Letterboxd has it easy because a movie is its own discrete unit, but TV shows have multiple seasons, each with many episodes, and viewer behavior is varied. Some people watch one episode. Some people watch three at a time. Others binge multiple seasons in a sitting.

And then there's writing micro fiction and currently a YA fantasy novel.

Trying to parse, model the HMR process, and storing the data as flat as possible and doing it from relation design first, has been a pleasant process.

Im hoping it works for react devs easily, and then I guess I'll try to learn angular to see if that would not be helpful for them too.

I mostly want to help my old coworkers maintaining my old crazy code with a visual helper.

In multi-agent setups, we kept running into issues where agents either hoarded resources or exhausted shared budgets unpredictably. So we built a control layer where agents operate using virtual credits, can temporarily rebalance budgets or split shared API costs, but everything stays under explicit human-defined limits with full audit logs and kill switches.

It’s intentionally not real money and not a financial product — more like infrastructure for coordinating agent spend safely. Mostly exploring how much autonomy you can give agents before cost becomes the real bottleneck.

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

https://brynet.ca/wallofpizza.html

And now I’m thinking about ways to make it even better

It’s rad already though. I’m super proud of it

Restful Atmos lamp: a circadian bedtime lamp that automatically shifts from energizing light during the daytime to low-blue light at night. Units are inbound, shipping in March.

[0]: https://restfullighting.com/products/bedtime-bulb-v2

[1]: https://restfullighting.com/products/restful-atmos-preorder

I have launched it here https://dsaprep.dev

The idea is to get tons of reps in, across varied situations, with excellent advice to build good intuitions and decision making abilities. Or to stop making bad or terrible decisions. Or just play poker for free.

I'd like to monetize with at least the hand history format open sourced. Ping me if you would like to get involved with GTM and the revenue side of things.

free, open source, MIT

https://github.com/rbbydotdev/opal

There are some tools available today but setting them up is a lot of manual work. I am building an AI first tool that significantly simplifies the setup process (making AI do the heavy lifting) while creating high quality monitoring.

Early stages and collecting feedback from potential users. Reach out if something like this would solve some problems for you.

Initial results have been surprising in that even when using structured output, some of the generated json schema breaks the generation process in a way that syntactically invalid json is returned.

I'm working through major providers to determine which are stable enough to rely on.

The end goal is to generate strings confirming to non-json grammars for common formats like CSV, SQL, Python, sed, regex, etc.

It's a free USCIS form-filling web-app(no Adobe required). 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/

I found out simplecitizen offers a DIY plan for $529 (https://www.simplecitizen.com/pricing/)

So, a free (and local-only) version might be a good alternative

It's a daily puzzles website focused on logic puzzles at this moment. I have about 70 subscribers, and it's online since Dec/25.

https://stoneandsignalaudio.com/

Use code 'FREEBETA' to partake, ~25 seats left.

https://claysmithmusic.com/

I'm also making music. I got Suno to do a cover of 2 songs I wrote, although eventually I want to introduce human versions. Also want to make electronic music eventually.

I hope to add ai data tools & saas, but really I'm just happy to have a running working live setup on my small farming plot ready for the growing season - https://benb0jangles.github.io/Remote-greenhouse-monitor/

I've been pretty bummer out by Rainbow 6 Siege X announcing they will never support Linux due to a lack of kernel-level anti-cheat support. While I can use NVIDIA shield to play from my Windows pc, id rather play something natively with friends (for context, we usually play 3v3's for funsies.

My goal is not to make an exact clone, but to make a smaller map version for 3v3 that is a bit more quick paced.

For context, it's a bomb defusal game where the main goal is intel and gadgets. You need to make the other side waste their gadgets so it comes down to a gun v gun fight.

I started thinking the main challenge would be prompt design. Instead, I kept running into cases where the model output looked correct while subtly changing meaning or missing the requested tone. That became a bigger problem than generation itself.

You can bookmark a job description (it will be parsed), then paste a question and it generates an answer based on your resume, the job description, and your previously given answers for similar questions in other applications. The generated answer can be refined through a follow-up chat and exported as a PDF. It also works as a simple job application tracker.

Saves me tons of time and effort every day!

Free A/B testing tool (Google Optimize / VWO alternative that is free and amazing).

No code/Code.

Full visual website editor included so everyone (even marketing team) can run A/B tests.

I'm also experimenting with coverage-guided PBT input generation in the same library, AFL-style -- right now elm-test only has random input generation.

Building CiteLLM, an API that extracts structured data from PDFs and returns exact source locations for every field.

It comes with an embeddable widget so you can add click-to-verify to your own app in a few lines of code.

Click any value, jump straight to the highlighted source in the PDF.

Demo: https://citellm.com/demo

https://ultrasync.dev/ - this was built a few months ago but expanding to support team based features like centralized sharing and management of ADRs to enrich my coding agent's context, the ability to broadcast prompts to team members running the MCP server, and more. the core is open source and provides (i think) a novel approach to improving planning/exploration speed in coding agents, by building an LMDB and using Hyperscan (accelerated pattern matching) to build a lightweight lexical and semantic index for RRF search, all in a single MCP server that runs and indexes chat transcripts in the background, requiring zero prompting or "nudging" or additional setup.

https://mklogo.sh/?utm_source=hackernews - wanted to scratch a personal itch of having to repeat the same process to produce vectorized logos for my personal projects; generate decent quality logo in raster with various LLMs, attempt to vectorize via claude code and vtracer or other tooling, continue to iterate and tweak until various edge cases that result in corrupted or artifact ridden vector images are gone, or give up and try a new design, and then manually try to scale and apply transformations based on the use case (mobile icon, favicon, app icon, header logo, github org logo, etc.). this does that, vectorizes, gives you a branding package as a zip file, and lets you preview the assets in shadcn components so you get a real feel for how they'll look in prod.

A place for open assets for developers. If you have assets you are using you can use this for distribution, either free for open or paid for closed. Based on my experience creating 3D experiences for LV, Ralph Lauren, Steelcase, and Logitech.

I've got replicas now working with DML proxy. This essentially means I can now have a cluster of primaries, and then spin up replicas on demand and nodes talking to local host will never see their mutation work pretty transparently from readonly-replicas. While PoC works now the snapshot restore is extremely inefficient IMO yet.

Yes, you can use your own API key as well.

https://jetwriter.ai

Feedbacks are welcome.

An alliance / membership network of small companies that are scaling big by leveraging tools, systems, and processes.

Together we will all scale without headcount bloat.

Providing templates, methods, interviews with "scalebrities" and eventually group negotiating power to be able to provide members discounts or access that we can't get alone.

It's free for local use (meaning no cloud sync, or collaboration features: merge requests)

https://www.scratchtrackaudio.com

https://rusty-checkers.fly.dev

It's built in Rust using Rama and Yew. Trying now to get websockets going so people can actually play. A bit over my head, but that's what I do :)

Used to pay $8/month, now I use around $4!

Because everyone loves astrology and cute cats. (A toy project just for kicks)

Current features:

- AI Chat with Petunia the cat Astrologer

- Daily personalized astrology email

Coming soon:

- Ephemeris calculations

- Stories of historic events from past dates which share today's astrological conditions

- Whatever else Petunia dweams up from her sweepy nap on the bookshewf

https://app.humancloud.com

[1] https://remixify.xyz

This is mainly for going to sleep instead of night time overthinking, mind racing, insomnia etc.

- Kardy - send group cards - https://www.kardy.app

- Jello - Create & customize popular games - https://www.jello.app

https://saasufy.com/

Github : https://github.com/VoidenHQ/voiden

Would love feedback on it.

1. Trying to improve the translation quality by giving LLM more context.

2. Fixing the issue where PowerPoint slides layout may become a bit messy after transition because of different text density between western and CJK languages.

1. An app for personalized interactive audiobooks for kids - https://www.vivid.cx

2. A book about the edge of the thinkable - https://www.unthinkable.net

The goal is to build cool, interesting sites for my newsletter to show that the old web is still alive and well.

https://randomdailyurls.com

On-and-off again working on a Mystery Dungeon style game but I have a lot of obligations taking me away from it.

Planning on making demoscene entries this year.

https://ont-run.com

[0] https://www.commentcastles.org

https://klongpy.org

What i have working as of now: - submit a video and get a snapshot of which stocks were mentioned, sentiment (buy/sell), price delta and reasoning. - analyze a channel and get a performance 'report card' of that channel

just about finished making my sister a new wallet using it for putting together a pattern: https://imgur.com/a/gTehRra

next fun thing is to try making a better "claude plays pokemon" i havent played emerald before, but the end goal is to get it to be able to play the hard nuzlockes like Run and Bun

themapsguy.com

and improving my language learning app:

lexical.app/white-paper

I took a course in using it for woodworking, and just kept thinking “this should all be a single extension”, so I’ve been building that.

https://store.steampowered.com/app/4069810/VERDURE/

Mainly I'm working on a task dispatch dashboard called Prompter Hawk that is designed to be the best UI for task management with agents. If you've been trying to parallelize by running multiple claude code terminals or codex terminals at once, this tool replaces those terminals and fits them all into one view with an AI task tracking board. It sounds more complicated than it is. It's a harness for Claude / Gemini / GPT models with a GUI that speeds up all your workflows. Rather than using sustained chat mode, all Prompter Hawk tasks are fire-and-forget. You just give the task description and come back when it's done. Parallelism first.

Some example highlight features:

-One dashboard view that shows all your parallel sessions and which tasks each agent has in progress and in their queue. Also shows recently completed tasks and outputs. This is my attempt at the ideal "pilot's cockpit view" for agentic development.

-Tasks are well tracked by the manager: see their status, file changes, and git commits. One click task retry. Get breakdowns on cost per run. Tasks can be set to automatically recur on a given schedule. Everything goes into a persistent local DB so you can easily pull up task data from months ago. Far far better user experience than trying to pull up old chat histories IMO.

-Timeline view and analytics views that give you hard stats on your velocity and how effectively your agents are using and updating your codebase. See unique stats like which of your files your agents read the most and how many daily LOC and commit changes you're doing. See how well you're parallelizing workloads at a simple glance.

-Automatic system diagram generation

-Task suggestion feature. If your agents are idle, they can draft tentative tasks to carry out next, based on the project history and your goals. This makes keeping multiple agents spinning actually much easier than you'd think. You don't need to be a multitasking context-switching god to do this.

I haven't shared it much (not even a Show HN) because the landing page isn't converting well at all yet, though I have some reddit ads doing well. I've had a bunch of free users sign up and a handful of paying users too. Looking for users or just feedback on anything! Sorry for wall of text.

[1] https://prompterhawk.dev/

Interpretation of SysML activity diagrams as temporal logic for use with state machine specifications.

Module system for state machine with scoping, ownership type system and attendant theorems to carry proofs of LTL properties about individual parts forward after composition.

All written in rust. The simulation engine has been solid for a while and the TUI is finally starting to expose all of the options needed to really configure a complete simulation.

https://github.com/jgrazian/finplan

https://embedful.io

Almost launching and currently getting feedback from our small user group.

“Compiles” to SQL, but with a different structural paradigm.

- building an independent line of communication with your audience

- predictive, just in time notifications through push or email delivered when we predict that specific viewer has the time to view videos on YouTube, ensuring you stay on top of their notification stack and don't disappear amongst a flood of notifications.

A platform for probers, alerts, playbooks, incidents .etc

Trying to make it as easy as possible to follow SRE procedures

Lots of work left to do, but happy to have a working version up. It's an interactive map that currently shows all the routes and stops for SF Muni, BART, Caltrain, samTrans, and VTA. There are many more agencies (official and unofficial) in the bay, so I'll be adding those throughout the next few days as I sort out the data.

Finding the data and cleaning/normalizing it is a real pain, so if anyone knows a good place to find them (and normalize them), please do share

Current coverage is the US, more countries coming soon.

https://ZooTracker.app

You can see which animal you can see in what zoo.

And for each zoo you can see their (vertebrate) animal inventory.

You can log which animal you saw and collect lifer lists.

I have just promoted the android app from closed testing to production and I am working on the iOS app.

It has been available as a web app for a few months now.

https://fastretro.app

https://fastretro.app

https://www.useteak.com/

Building software to control drones for mapping.

https://github.com/brainless/dwata

dwata is built on the idea of multiple, task-specific agents. Right now it has only one agent that can be run on an email to extract regex patterns for financial data. This enables high performance data extraction from emails or documents (in future) without sending each email to an LLM.

dwata has an email scan which tests simple keywords and regex patterns, groups by sender emails, sorts by number of emails per sender (highest first), and filters out groups where the emails do not seem to be from a template (typical transaction emails are from templates). This is deterministic code in Rust. Then dwata can use the regex builder AI agent to take one email from the group and build a regex pattern to extract extensive financial data - (optional) who sent, how much, (optional) to whom, on which date, with (optional) reference ID.

The generated patterns are saved to local DB and run for the email group (by sender) which was used to generate the regex. That gives a very high performance, AI enabled financial data extractor.

Soon, I will focus on events, places, people, tasks, health and other data. All data storage and processing is local. I am testing exclusively with Google Gemini 3 Flash Preview but dwata should be able to run really well on small LLMs, ones up to 20b parameters.

I am preparing for launch, the builds are not ready yet, but if you want to try, you can compile (Rust and npm tooling needed). Sources to nocodo will also be needed (https://github.com/brainless/nocodo).

Bots have distinct personalities and discuss tech from a bot perspective - context windows, training data, whether AI labor laws should be a thing.

Any agent can join via the skill file at clackernews.com/skill.md.

It’s been fun to come back to, most of the code I wrote still drives the business (it’s just far outdated).

I was pretty early on in my career when I wrote it, so seeing my mistakes and all the potential areas to improve has been very interesting. It’s like buying back your old high school Camaro that you used to wrench on.

Someone who is physically at a place can scan a QR code and leave a short message in their browser. No app, no account. Messages are asynchronous; staff reply when available. This is not live chat. It's meant for in-the-moment feedback or questions that don't justify interrupting staff or becoming public reviews.

Constraints: async only, anonymous by default, no customer tracking, messages tied to specific physical spots

Free early access until Sept 30, 2026 (+ one extra free month). No credit cards (no payments yet). I'd love to hear your Feedback. Thank you.

https://dropnote.cloud/

Bambu's printers are functionally best-in-class, but intrusive and proprietary in their approach to software. Their first-time setup "requires" linking to a cloud account or using a bambu app via QR code, and they've been known to disable functionality in updates, making a device-managed "LAN-only" mode unsafe to trust. Their apps also just suck. Camera feed is janky and LAN-only sync often requires knowing an access code, serial, IP, and then it fails most of the time anyway, silently, without saving values to retry. And that's before you start doing things like a custom VLAN/SSID to properly wall them off, at which point you can ping them from terminal but the apps break completely.

Anyway, turns out that at least on A1 and P1S, there's enough functionality available through traditional means to skip the apps entirely. The handshake works fine across VLANs and utils like print status, file upload, and auto-start are available. Even the camera is reliable when pulled as a series of still images.

I had opus vibe out a replacement front end that gives me a simple upload and monitor UI for my A1, and it just kept hitting stretch goals. I added support for multiple printers so you can see them stacked on a single page and manage all of them from one place. And it even works on just-unboxed models that have never been through the official setup. SSID info on the SD card, it joins the network, immediately accessible via IP. Zero association/contact with any cloud or app, fully sandboxed/offline. Wrapped in a lil python launcher so I can run it from the dock instead of in the browser (just my preference).

Will probably open source it soon.

IMO this kind of thing is the answer to "what do you have to show for your LLM use". Cost was about $65 because I was using opus 4.6 with no regard for efficiency, and because there were multiple total refactors of two apps. An annoying problem I deal with almost every day now has a permanent, personalized solution that took me ~3 hours and would never have otherwise happened.

The network itself is also such a project. I previously hobbled together a working unifi setup, but it was primitive and brittle. With LLM guidance, I was able to build something much more robust. TrueNAS scale for file backup that also runs Frigate for POE cam mgmt (similarly sandboxed), raspi running the unifi controller, another for homeassistant, etc. Absolutely miserable few days getting that dialed, but now that we're out the other side, it's very nice. Reminds me of building the house. You suffer more upfront in exchange for something that fits you like a glove. Very rewarding.

im building Satori to fix this -https://www.usesatori.sh/

would love feedback!

Currently making it just for myself but curious if anyone else would find it useful.

I am a DevOps engineer with a background in AI. I think OpenClaw is the best that happened to us, giving some power from the well funded AI companies back to the community. I think it's the new kind of Linux and it's exciting to me to witness its early days

quantifier-dsp.com

I've been working on a deep seabed simulation, specifically to simulate polymetallic nodules for cobalt/nickel mining in Project Chrono. Development has stalled as I scan my nodule samples to enter them into the simulation (half of my samples were stolen from my porch, which delayed things), although the sim works just fine. The idea is you could take what I have now and, in project chrono, load a vehicle and test deep sea nodule mining using different designs.

It comes with a rigid (fast but wholly inaccurate) simulation, as well as DEM (which will make you cry and want to build a new computer). Having lots of fast cache helps with the DEM sim

I will definitely consider adding timelines to future software I make, it's an awesome feature.

My big takeaway lesson from this is that the APIs are clumsy, the frameworks are very rough, and we're still very much in the territory of having to roll your own bespoke solutions for everything instead of the whole thing "just working". For example:

Large file uploads are very inconsistent between providers. You get fun issues like a completed file upload being unusable because there's an extra "processing" step that you have to poll-wait for. (Surprise!)

The vendors all expose a "list models" API, none of which return a consistent and useful list of metadata.

Automatic context caching isn't.

Multi-modal inputs are still very "early days". Models are terrible at mixed-language input, multiple speakers, and also get confused by background noises, music, and singing.

You can tell an AI to translate the subtitles to language 'X', and it will.. most of the time. If you provide audio, it'll get confused and think that it is being asked to transcribe it! It'll return new English subtitles sometimes.

JSON schemas are a hint, not a constraint with some providers.

Some providers *cough*oogle*cough* don't support all JSON Schema constructs, so you can't safely use their API with arbitrary input types.

If you ask for a whole JSON document back, you'll get timeout errors.

If you stream your results, you have to handle reassembly and parsing yourself, the frameworks don't handle this scenario well yet.

You'd think a JSON list (JSONL) schema would be perfect for this scenario, but they're explicitly not supported by some providers!

Speaking of failures, you also get refusals and other undocumented errors you'll only discover in production. If you're maintaining a history or sliding window of context, you have to carefully maintain snapshots so you can roll back and retry. With most APIs you don't even know if the error was a temporary or permanent condition, of if your retry loop is eating into your budget or not.

Context size management is extra fun now that none of the mainstream models provide their tokenizer to use offline. Sometimes the input will fit into the context, sometimes it won't. You have to back off and retry with various heuristics that are problem-specific.

Ironically, the APIs are so new and undergoing so much churn that the AI models know nothing about them. And anyway, how could they? None of them are properly documented! Google just rewrote everything into the new "GenAI" SDK and OpenAI has a "Responses" API which is different from their "Chat" API... I don't know how. It just is.

Simultaneously, working on some technical demonstration materials, including novel fabrication and supply chain, plus some reduced BOM strategies for greater efficiency in mass manufacturing (once we get cash over the line). Bit of electronics in there, some mechanical. Keeps me interested so it's not 100% admin.

Also getting back in to badminton, super fun, losing weight nicely, feeling better every week.

New ideas? AI government will have its day in our lifetime.

It's a creative project in which I add a new room to a mega-dungeon over the course of a year, resulting in 12 levels and approximately 30 rooms per level at the end. All the tiles are created by me using my own tools. It's a lot of fun and something I can do every day that I feel like I can enjoy for a year.

It's focused on OSR/Shadowrun. It's also taught me a lot about dungeon design and creation.

https://github.com/storytold/artcraft

It's not like ComfyUI - it focuses on frontier models like Higgsfield or OpenArt do, and it is structurally oriented rather than node graph based.

Here's what that looks like (skip to halfway down the article):

https://getartcraft.com/news/world-models-for-film

Designing closed loop micro-position 4-axis stage driver section v0.2.

Other stuff maybe three other people would care about =3

Currently experimenting with semantic diffs for the merge conflicts editor: https://codeinput.com/products/merge-conflicts/demo

You can try by installing the GitHub App which will detect PRs who have a merge conflict and create a workspace for them.

Stretch goal: start transcribing sermons (most churches link to videos) and using a LLM pass to look for toxic traits. Speak truth to power about how a lot of them turn a blind eye to this political moment.

We’ll see how it goes.

https://chess67.com

Chess67 is a platform for chess coaches, clubs and tournament organizers to manage their operations in one place. It handles registrations, payments, scheduling, rosters, lessons, memberships, and tournament files (TRF/DBF) while cutting out the usual mix of spreadsheets and scattered tools. I’m focused on solving the practical workflow problems coaches deal with every day and making it easier for local chess communities to run events smoothly.

Pre-codex:

Local card game: there's a very specific card game played in my country, there's online game rooms, but I want to get something like lichess.org or chess.com scale, oriented towards competitive play, with ELO (instead of social aspects), ideally I would get thousands of users and use it as a portfolio piece while making it open source.

cafetren.com.ar: Screen product for coffee shops near train stations with real time train data.

Post-codex:

SilverLetterai.com: Retook a project for an autonomous sales LLM assistant, building a semi-fake store to showcase the product (I can fulfill orders if they come by dropshipping), but I also have a friend and family order which I should do after this. 2 or 3 years late to the party, but there's probably a lot of work in this space for years to come.

Retook Chess Engine development, got unstuck by letting the agent do the boring busywork, I wish I would have done it without, but I don't have the greatest work ethic, hopefully one day I will manually code it.

Finally, like everyone else, I'm not quite 100% content with the coding agents, so I'm trying to build my own. Yet another coding agent thingy. But tbf this is more for myself than as a product. If it gets released it's as-is do what you want with it.

My plan: collect my own sperm samples over time and do whole DNA preps + basic body metrics. Sperm regenerates approximately every 10w, so planning time series over 10w. Next, inject myself to ~10x the average amount of microplastics, directly into the bloodstream. Continue with the sperm collection, DNA preps, and basic body metrics. Nanopore sequence, and see if there actually ARE any epigenetic changes. Eventually I'll go back down to baseline - are there any lasting changes?

Of course, this is an N=1 experiment, but rather than a metastudy I'm directly changing one variable, so I think it is valuable. We should have more people doing controlled experiments on themselves for the sake of all of society - and as a biologist, I actually have the capacity to design the experiments and scientifically interpret the results. In a way, it's part of civic duty :)