Ask HN: What Are You Working On? (April 2026)

After the LiteLLM supply chain hack last month I feel pretty good about that choice. Your LLM gateway holds every provider API key you have. That probably shouldn’t be a pip install.

Worked on API gateway for a major retailer and incorporated my learnings into this platform.

Open source engine take a look!

I'm presenting at LinuxFest this year, so I'm currently in the process of wrapping up my slides for that. They're turning out ok; I have had to resist the urge to have AI write them for me, since they has a tendency to make everything feel soulless.

I maintain a fork of the main MiSTer executable [1] because of some disagreements with how Sorg runs the project and because I want to reduce the risk of saves being corrupted. Now I'm trying to come up with an automated way to monkey-patch the upstream changes so I can apply my changes on top. I have been experimenting with putting something like Claude in a Github Actions to handle this, but I haven't nailed down anything I'm super happy with.

I have been on a quest to find the source code for the old Digital Research Concurrent DOS. It's taken a few turns and I've been blogging about it: https://blog.tombert.com/Posts/Technical/2026/03-March/The-Q...

Because I have to accept the fact that I may be unsuccessful with finding the source to Concurrent DOS, I have been learning how to do reverse engineering with Ghidra if I ever want to see even a facsimile of the source code. Once I get competent enough with that I want to play with the MCP for Ghidra.

I have grown tired of people committing AI generated code, so I've been working on a library (written by my own fingers) that allows you to "assume" certain functions exist and have AI generate it for you, and use aggressive memoization to avoid it being too expensive. I'm working out the kinks and trying to make it more modular and flexible and deterministic, but I think it would be kind of neat. It's Opus doing the real work, of course, but for example I told it to `assume` that there existed a symbolic differentiator function that took in a string and did a derivative with respect to x, and then another function that could take in a function that took in a string of a polynomial and made a regular function out of this; basically assuming a very light version of Mathematica.

I think that's basically it for now.

[1] https://github.com/Tombert/Main_MiSSus

The whole thing started because my wife couldn't get into the official Audiobookshelf iOS TestFlight beta. Her exact words were "I cannot live without audiobooks." I'm a backend dev, never touched iOS or Swift before, but how hard could it be? (It was quite hard.)

About a year in now. CarPlay, offline downloads with background sync, Cloudflare Access support for tunneled servers, sleep timers that create bookmarks so that you can remember where you were the next morning. Currently working on podcast support. Solo project - no tracking, no accounts, just talks directly to your server.

If you self-host your audiobooks and have an iPhone, give it a shot: https://apps.apple.com/app/soundleaf/id6738428638

I'm an avid weekend golfer trying to improve. I noticed a lot of negative self-talk after a poor streak on the course and picked up "How Champions Think" as recommended by r/golf. In a chapter on optimism Bob Rotella shares a story about how, before he won the Master's for the first time, Seve's friend made him a tape of a fake news broadcast of him winning it that he listened to obsessively leading up to the tournament. Victorious lets users generate similar visualizations for any big moment (sports, performances, public speaking, job interviews, hard conversations...).