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

Posted by david927 16 hours ago

What are you working on? What have you been curious about lately?

161 points | 501 commentspage 9

trwhite 1 hour ago|

I'm working on "See My Spending" - a web app that gives you better insight into where your money's going. It provides real, actionable insights without the use of generative AI. Currently preparing for a launch for UK customers only

robgough 8 hours ago||

At the beginning of this year I was fractional CTO at a small b2b startup and was surprised by how what I considered to be the basic requirements of a customer support platform was split across multiple tools and all priced per-user. This essentially forced a small team to limit who in the org had visibility into what was being communicated to customers.

I set out to solve that with https://stayupfront.com and now that the build is in a good place i’m finally having to bite the bullet and learn marketing. I’m also realising that all the advice to start with marketing first was probably good advice!

aadyachinubhai 2 hours ago||

https://github.com/aadya940/scikit-verify

Trace Python+NumPy programs into SymPy formulas and check them symbolically, across all branches

mdbug 1 hour ago||

https://jeden.day

A habit tracker that allows for more flexible goals than daily ones, such as N times per period.

spankalee 3 hours ago||

I'm working on a new programming language for WebAssembly GC, called Zena: https://zena-lang.dev/

It's like a fixed up TypeScript specifically tailored for Wasm GC that can produce very small and fast binaries, but also has features like pattern matching, ownership / borrow checking for resources, pipeline, tail call elimination, multi-value returns, direct WASI component integration and a lot more.

NetOpWibby 3 hours ago|

This looks pretty cool! I love TypeScript. Although for my server-side apps, I just use Deno and compile my executable that way.

I have pondered cloning Caddy in TypeScript just because...would Zena be capable of this?

tducret 1 hour ago||

https://apps.ducret.dev/har-analyzer/ (demo video in the page)

I'm building HAR Analyzer: a native macOS app for searching and inspecting HAR files locally, without uploading sensitive traffic to the cloud.

Like Network Dev Tools but with a better UX.

LamGC 6 hours ago||

I am currently developing a great encryption tool for long-term highly confidential files in the post quantum era, but I am still concerned about any issues with it, so I am personally using it for a period of time to prove that this tool can be released as open source! (I plan to personally use it to encrypt and backup some of my GPG keys offline, then burn them onto M-DISC discs and store them in a bank safe)

stackghost 6 hours ago|

> I plan to personally use it to encrypt and backup some of my GPG keys offline, then burn them onto M-DISC discs and store them in a bank safe

Won’t that make it difficult to rotate your keys?

Keys should be something you don’t hesitate to throw away and rotate at the first hint of a compromise.

LamGC 6 hours ago|||

My GPG key is renewed expire date every 2 years, and the main key and sub key are separated (the main key is stored offline, and the sub key is on YubiKey). The private key does not need to be updated regularly, so it can be directly encrypted and saved in the backup.

I think this is a good balance point, and I estimate that if we want to rotate the keys completely in the future, we should start using new key algorithms. (I estimate it will take another 5-8 years)

LamGC 5 hours ago|||

> Keys should be something you don’t hesitate to throw away and rotate at the first hint of a compromise.

Yes, when certain situations arise, it is necessary to rotate the keys as soon as possible, but in the absence of such situations, losing key backups can be catastrophic.

cutwaterflow 7 hours ago||

I’m building https://www.freefocusgames.com, a free collection of browser-based attention and working-memory exercises, including Dual N-Back, Stroop, Schulte tables, attention-control exercises, and memory games.

I started it because I wanted simple brain-training exercises that didn’t require an account, app install, or subscription.

It’s also open source: https://github.com/loethen/freefocusgames

aprct 7 hours ago||

Just a heads up: The word memory test is case sensitive. My mobile phone defaults to capitalizing the first letter of those inputs, so the result was 0% accuracy despite technically remembering the words. I don’t know if case sensitivity is deliberate, but I chose not finish the test.

cutwaterflow 3 hours ago||

Sorrry about that! It definitely wasn't intentional - i've just fixed it. Thanks a lot for the heads-up!

switchbak 6 hours ago||

That takes me back. I made an unreleased Android dual-N-back game in the late 2000’s, in Scala no less. I got a little bummed when the big claims of the N back training failed to be replicated though.

cutwaterflow 3 hours ago||

Scala on Android in the late 2000s? That’s seriously hardcore! Back then I didn’t even know how to write code yet. That definitely brings back memories of an earlier era.

xkam 2 hours ago||

https://multicoder.dev/

One UI for every coding agent - VS Code extension (GUI) to launch and work with any ACP-compatible agent harness. Supports Claude Code, Codex, OpenCode, Qwen and many others - https://agentclientprotocol.com/get-started/registry . Does not require any new subscription or user ID - this is still handled by the harness, so you can use existing Claude Code or Codex subscription. Multicoder just gives you a high-quality GUI to work with any of them without mental switching cost.

ponyous 15 hours ago|

https://grandpacad.com - AI modeling for 3D printing Been at it for about year and a half.

Really exciting stuff is happening literally every month, because underlying models are getting better and better. When I started it was pretty basic: “make a cube with a hole through it”. Now it’s at the point “make a raspberry pi 4 case” and the agent searches, builds, verifies…

What surprised me in this process is how little meaning AI benchmarks have. Pareto frontier for my use case looks completely different than any other benchmark portrays.

jamesponddotco 15 hours ago|

You mentioned benchmarks don’t mean much in this area. What are the best models you found in this area?

Personally, I’ve had the most success with GPT models using MCP servers for this task.

ponyous 2 hours ago|||

Surprisingly Gemini models. Spatial understanding seems to be the best for the price and speed.

dr_kiszonka 8 hours ago|||

Same here. 5.5 would sometimes oneshot relatively complex models even without MCP. Opus, however, even with Fusion 360 MCP access would usually make comically bad models. It seemed that Opus tried to reason through the problem while GPT would come up with a general plan of using whatever tools were available to solve the problem. In chat, it would use the available Python and trimesh. It was also the only time I had ChatGPT work on something for over 30 min without giving up.

More comments...