Top
Best
New

Posted by david927 1 day ago

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

What are you working on? Any new ideas that you're thinking about?
258 points | 968 commentspage 15
pvillano 11 hours ago|
I made sdf2stl.saej.in

You can use it design 3D objects with mathematics and bring them into the real world with 3D printing.

You can use it to create 3D models that are impossible to create with CAD, CSG, sculpting, or mesh-based tools.

You can build on a decade of community SDF development on Shadertoy because sdf2stl uses the same language.

A few days ago I may have become the first person to 3D print the equation x^4+y^4+z^4-x^2-y^-z^2+.4=0 (Goursat's surface) https://www.printables.com/model/1713835-goursats-surface

Please try it out and tell me what you think.

https://sdf2stl.saej.in

davidpolberger 1 day ago||
I'm working on turning our statically-typed formula engine -- that we use for Calcapp, our app builder -- into a real hosted solution (as well as a library). I discussed it in July last year (https://news.ycombinator.com/item?id=44702833#44704642) and have been working full-time on the project since the beginning of the year.

I figured "I already have a battle-tested solution, I just need to make it modern and spiffy, build a website for it and see if there's any interest -- in the age of Claude Code, this should be fast work!"

Wrong. Taking an internal library and offering it to others -- complete with documentation and modern tooling -- is an immense project, even with the help of AI agents.

Is there a market for a "formula engine in a box"? I don't know. But I also didn't know whether there would be a market for Calcapp either, and that has supported me working full-time for the past seven years. So I'm willing to take another chance.

serickson 10 hours ago||
https://casedaemon.com/

CaseDaemon: automated intake handling for immigration lawyers. Given a USCIS form to fill out and a set of documents and information, CaseDaemon automatically fills out the form with what it has and prompts the client (directly or via the user) for additional documentation or information needed to finish the application. Takes out a bunch of the back-and-forth between lawyer and client, and the busy-work of the lawyer mapping data to form and tracking requirements.

The product will be ready to use in a few weeks, but take a look at our homepage in the meanwhile, curious what people think!

ryanczak 7 hours ago||
refactoring memory management in https://github.com/ryanczak/daemoneye to better support continuous operation over long time horizons where state of monitored services/things drifts and knowledge becomes stale over time.
renegat0x0 10 hours ago||
I maintain databases of links

- h ttps://github.com/rumca-js/Internet-Places-Database - Internet places / YouTube channels

- h ttps://github.com/rumca-js/awesome-database-feeds - feeds / RSS locations

- h ttps://github.com/rumca-js/awesome-database-top - smaller database from above

- h ttps://github.com/rumca-js/awesome-database-awesomelists - links from 'awesome lists'

- h ttps://github.com/rumca-js/RSS-Link-Database-2026 - 2026 year link metadata

- h ttps://github.com/rumca-js/RSS-Link-Database-2025 - 2025 year link meta data

- h ttps://github.com/rumca-js/crawler-buddy - crawler engine

slater 10 hours ago|
is there a reason for the space in "h ttps" ?
srean 8 hours ago||
Probably to prevent the HN hug of death.
ecto 23 hours ago||
I'm building https://vcad.io, a free CAD application with an open source kernel (Rust/WASM)!
frostbyrne 1 day ago||
Greetings!

I've been working on something in the vein of a indie game for a little over a year now. It has been a passion project, but I'm starting to come around on showing it to people.

I am a big fan of Telltale style narrative games. I think Baldur's Gate 3 was the biggest revelation of this for me. Taking that branching dialogue and freedom of choice, and tacking it on to a fun combat system was just everything.

When text based GTRPGs started popping up, I found it hard to connect with them stylistically. I found that I needed the multimodal stimulus of visuals and audio. This led me to start building something, and it ended up being somewhat of a cross between a Telltale game, a Visual novel, and a TTRPG.

Orpheus (https://orpheus.gg) is a fully on-the-fly generated tabletop simulator, with graphics, audio (TTS), and the freedom you can usually only find at a real TTRPG table. That means you can play a sci-fi, fantasy, or even a modern setting in your campaign. The assets are made for you as needed. It runs in your browser so nothing to install or tinker with.

Getting the harness right so the AI GM can stay coherent and organized has been the biggest challenge. It took a lot of iterations to get it to a point where it could understand the scenes it was building as the player changed them.

I've built it to be played with either a keyboard or a gamepad so you can play from your couch. You can switch between them as you feel like it. There is a 3D tabletop for combat, full character sheets, dice rolling, lore tracking. I want it to be dense.

Mostly, I’m looking for people who want to try it, break it, and tell me what feels magical, confusing, boring, or broken. My biggest roadblock currently is that asset generation is relatively expensive. I'm currently mulling over whether a playtest would allow for a BYOK setup so people could try playing as much as they'd like, or if I should add turn limits.

You can join the playtest waitlist at https://orpheus.gg/ -- and I just setup a discord (https://discord.gg/pychWyzf) that I will use for early playtests. (Just me right now! Come hang out!)

vunderba 1 day ago||
I've been the DM of a weekly campaign with the same group of friends for nearly a decade now. Over the past few years, I’ve seen a lot of attempts at AI meets D&D, and most of them suffer from fairly pedestrian puzzles and stories, and don’t really compare to what a decent, semi-competent human writer can come up with.

I'd love to see a more modern day attempt at something like Bioware's Neverwinter Nights - which was designed so that someone could create a campaign, and then the game would provide the behavior, pathfinding, assets, and everything else with a virtual (or human) DM behind the scenes. You could still tell a human-driven story, but the engine would do a lot of the heavy lifting.

frostbyrne 1 day ago||
I agree, there is no way to perfectly capture a real table with your friends. This is more an alternative for when life gets in the way of meeting every week.

I think a lot of those attempts you mentioned try and brute force the problem or trust the AI too much on what to generate.

A lot of the same problems that AI coding agents run into also apply to this problem. You have to really manage context (avoid sending a novel at the model) and enforce strict rules in the "engine". The hard part is world building that is consistent without railroading the player and forcing specific paths. I have an agent (for lack of a better term) that manages arcs across each tier. World arcs (nations, factions), player character arcs, NPC arcs, individual scene arcs, and location arcs (towns, cities, dungeons, etc). By prompting all of these as tight, individual arcs with flavor and context peppered in as needed, you end up with stuff that is more compelling. It has to be loose enough that you don't railroad the player. When you decline that NPC's quest, down the road that might have changed the overall arc for a town in a meaningful way.

I won't pretend that I've perfected anything but I have definitely noticed a spark in its writing and world building that I personally have really enjoyed.

vunderba 1 day ago||
Yeah I'd agree that I think AI can at least work provided you manage the context properly, multiple top-level files establishing consistent world state, all that jazz. KoboldAI and SillyTavern both do a pretty good job of maintaining internal consistency around longform interactive fiction.

OTOH, that means that the underlying story is that much more important. I think a lot of people mistake coherence for novelty. Biggest offender is puzzles - oh god do LLMs absolutely blow dire wolf chunks at coming up with organic and interesting puzzles.

blinkbat 1 day ago|||
Why not generate some asset libraries to help with some of the rote generation? You could theoretically serve the same asset for a pack of rats to multiple campaigns.
frostbyrne 1 day ago||
Yes! I am doing a lot of this where it won't break the illusion. Not everyone needs a unique innkeeper generated in every town, but I want to avoid that "Officer Jenny" effect like in Pokemon where she looks the same in every town they visit.

I have a private vs public flag for assets that I'm considering more unique or sensitive, at the AI GM's discretion. I'm using embeddings from there to try and parse if an asset already exists in the public pool or not, and reuse it if possible. The thinking is that eventually I will have pretty decent asset coverage on most standard campaigns. I can't account for people going way off book though.

I have an asset pipeline that tries to determine player intent and pre-generate assets before they're needed. That way we can attempt to hide the "load screens" like retro games did with elevators. I have a kind of sliding scale for player coherency, and if the player has too many "misses" on the pre-generation pipeline it will increase its requirements for when it starts generating.

I may have wildly over-engineered this but I love it. =)

iugtmkbdfil834 1 day ago||
Super cool project and I love to encourage this as I personally think llms in games have not been utilized to their potential quite yet.
rambleraptor 1 day ago||
I’ve been working on an OSS backend-in-a-box called [aepbase](https://aepbase.io/).

For the past few years, a group of us from Google, Microsoft, GM, IBM, Roblox, Rubrik + more have been working on a design standard for APIs called [AEP](https://www.aep.dev). The goal is twofold: learn from our companies mistakes around APIs and enable better tooling with less configuration.

We’re at a point where AEP-compliant APIs get a resource-oriented CLI, MCP server, full UI, and Terraform provider for near-zero configuration.

Aepbase has been my way to tie the whole ecosystem together. You run a single binary and define the schema for a resource with one API call. Now, you’ve got a full set of CRUD APIs and support for CLI/TF/MCP/UI. After one API call.

It’s a really cool way to tie together all of the work AEP has been doing.

Love to hear HN’s opinions on all of this. We’re still trying to figure out the best way to sell people on AEP.

umairyousif2280 5 hours ago||
I recently built this project for AMD AI Hackathon. At it's core, it's a local first debugging tool thats powered by Qwen 3.5:9b.

You give it a repo and an error log and It finds the root cause, suggests a fix, shows the bug propagation and the files it went through. The fix it suggests is in a diff format so its very easy to apply the fix. Here's the link to the repository if anyone wants to try it.

Any type of feedback is welcome!

https://github.com/umairyousif239/AxionSys

skataz 5 hours ago|
I am working on collecting.cash, a webapp that helps with sending payment reminders. Can be used for reminding members of a family plan (mobile, or streaming, etc.) when a payment is due.
More comments...