alt Hacker NewsAsk HN: What are you working on? (August 2026)
What are you working on? What have you been curious about lately?
Comments
I'm working on https://sunware.ai/ to automate software work like GitHub Actions.
An agentic marketplace where people can buy business services from inside their claude code: https://talkshi.com
dotIPA, an iOS app build size inspector that runs locally on your macOS [$4.99] Track app size growth over time, inspect contents, spot duplication and size bloat and more. https://apps.apple.com/us/app/dotipa/id6742254881
Building an AI agent that creates promotional content for social media, in a way that deeply resonates with your target audience. I tried using Claude and ChatGPT first, and they could give me a few good posts, but not a reliable stream of them. After 10–15 outputs, the angles converged, the language repeated, and everything became too salesy. I was still providing business context, finding angles, tracking previous posts.
It was clear to me that a marketing harness was needed around the LLMs for this purpose, and so Feral was born.
https://runnit.io - creative team work/project management system
I've rapidly been clearing some of the smaller items off of my ideas list while procrastinating on my PhD:
- nichromecast - TUI/CLI Chromecast sender with for local files and YouTube URLs - https://github.com/evidlo/nichromecast
- syncthingtui - TUI for administering Syncthing on headless machines - https://github.com/Evidlo/syncthingtui
- Need for Madness Web - WebGL port of an old Java racing/derby game - https://radicalarchive.github.io/nfm/
- MdInvetory - Markdown-based personal inventory editor - https://github.com/evidlo/mdinventory
my own high speed train viewer (trainradar) that i use to further develop my software development skills. so far i'm happy with it and its performance, and it has forced me to learn how to scrape without being detected/banned
i'm using react on the front, pretty interesting how it manages states in the frontend and how to come up with new strategies to not trigger so manny re-renders of the UI, and on the back i use fastapi with postgresql and oauth2
I'm working on Pawvis, visual hand-tracking mouse & voice control that's open-source and fully local on your Mac (with optional handoff to Codex or Claude Code for complex computer use tasks).
We live in the future, and I wanted to use my computer the way it's been depicted in movies for decades. It turned out more capable than I'd expected and frankly really fun to use.
Homepage: https://pawvis.app
Nothing. I get no satisfaction out of side projects nowadays
g023 Code - https://github.com/g023/g023code a deepseek v4 flash harness built specifically for that model, that uses Ollama for its vision support, and uses the built in web search capability natively. Created only in Python to keep it easy to implement and use. Made to achieve cache efficiency, and keep context tight.
lazyslurm is a tui for managing slurm jobs on HPC clusters. It's kind of like lazygit or lazydocker. I've been chipping away at it for the last few months. Its built in rust w/ ratatui.
A new task manager to replace Taskwarrior
AI sense of humor benchmark: http://witbench.com/
I'm collecting ideas for my gnomes in Pakistan to work on while pretending to be deeply interested in yuppies' personal lives, lol
I had issues with reproducing some fuzz testing failures in CI locally; turns out that glibc, musl and macOS actually DO NOT return the same floating point values for certain trig/transcendental functions across all OS'es, so I decided to rectify that by spelunking a deep rabbit hole: https://github.com/pmarreck/random
In order to make it work, I had to re-implement some math, so I avoided floating-point altogether (IEEE754 is a personal pet peeve) and went with mantissa and exponent fixed-point decimal. Or should I say binary, because the "decimal place" is the number of bits, not the number of base-10 decimal places.
Apparently this is the only deterministic, cross-platform, cryptographically-secure procedural RNG that has many types of distributions (uniform (default), normal (Box-Muller), exponential, Poisson, log-normal, beta... all with configurable inputs AND which will plot the distribution for you in the terminal!) AND is guaranteed to give you the same values... Everywhere. (Because how the fuck has this not existed yet??) Implemented in both LuaJIT and Zig with C FFI. So there you go. MIT license. Enjoy your deterministic but still very random numbers, on every OS, identically.
(It sources the best-in-class recommended source per OS if you want true random. It will do that too.)
I'm working on an open source project to make drone delivery accessible across any industry not only the software, but also the drone hardware
Working on sauna & cold plunge app. https://sauna-assistant.com
An AI agent to convert camera roll shots into cinematic reels with narration + music. Each scene of the generated reel is editable by human.
I am working on agentic infrastructure for deploying very small production apps en masse.
In particular, I’m exploring what the infrastructure should look like when agents are the primary users: deployment, databases, domains, secrets, observability, and lifecycle management.
The goal is that everybody and their pet can deploy an app.
These units are called saasies, and the project is Saasie.
Forex analysis software based on waves https://dxodus.com
Some publishing for my kids, turn a pdf into a zine
Hypnolines! https://epixo.net/hypnolines.html
tirreno - security framework
building kubehub.io , build a private cloud with your own hardware, manage through kubernetes, most important easy to use.
kubernetes on prem setup take a lot of works, even if you setup the cluster there are tons of work wating for you to setup ingress/gateway/DNS/TLS/monitoring.
kubehub.io solves all those problems.
I'm creating a trading bot. Сreating my own agent for a local LLM to help me write code.
still trying to figure out how to find users for my offline model agent harness with self-customizability and an offline wikipedia reader/tracker. tell your friends if you can think of a meaningful audience profile. http://enough.support has details/links
Harness evals on private benchmarks. I haven’t found a good way to evaluate harness releases on realistic private tasks
Building a Web-based guessing game called „Guesstimate“ where the player needs to guess age, capacity and height of things. Includes friends leaderboard and player‘s statistics:
Keep hearing that management sucks so building a tool that helps new and experienced manager improve and grow.
Got laid off in June. Grinding leetcode.
a new agent/harness, mu. not ready yet but hope to let people play with it ~soon. its got very few novel ideas, but attempts to do the existing good ideas reasonably well and be as minimalist and composable and pragmatic as possible
It's a bookmarker tool.
I've been building a cicd + orchestration layer, and on-off hacking a 2d animation engine.
A few things outside of normal work that are deliberately different to be interesting to me:
- Ongoing work on an airline announcement engine that works with flight simulators
https://github.com/fearlessfrog/MSFS_Universal_Announcer
- A multiplayer board game for Discourse forum users that is not the popular Risk (tm).
https://github.com/fearlessfrog/discourse-not-risk
Both have a fair bit of traction, but is often the case, keen to do something new this month.
An open source, decentralized, interoperable marketplace backed by open-source SaaS for every vertical
I'm working on a Python rewrite of Anacreon: Reconstruction 4021.
update to my ai first desktop https://apps.apple.com/app/aifcc-ai-first-computer/id6782364...
Making rent as an open source developer.
Desperately trying to attract new monthly sponsors and people willing to buy me the occasional pizza with my terrible HTML skills. Is it working?
If any individuals, companies (or bitcoin millionaires) would like to help a long-time OpenBSD slacker, unslack, I'd really like to focus more of my time on open source development (and advocacy), rather than making rent. Feel free to contact me.
https://brynet.ca/wallofpizza.html
(Native SegWit): bc1qwe6zv0ezq4gzlea6tw45qhsn5kckheljn0krvt
I just released a small app called Fogpane: https://asok.github.io/fog-website/index.html?lang=en
The concept is intentionally simple: you open a photo, it gets covered by a layer of fog, and you use your finger to wipe the fog away and reveal the image underneath – a bit like drawing on a fogged-up window.
I’ve always wanted to publish something on the App Store, so I decided to use this as an experiment: how far can I get while letting AI write the entire app?
For full transparency, essentially all of the code was written by AI. I guided it, described what I wanted, tested the results, reported bugs, and made product/design decisions, but I didn’t really write the implementation myself.
For an app this small, I was surprised by how far this approach got me. It wasn’t completely hands-off – there was plenty of iteration, especially around making the fog and water droplets feel somewhat realistic – but it went from an idea to an actual app in the store.
I’m a software developer myself, which made the experiment particularly interesting. In many cases I understood what the AI was doing, but deliberately tried not to take over and write the code myself.
I installed Codex around three weeks ago. It’s hard to estimate exactly how much time I’ve spent on the app since then, but it’s probably been a couple dozen hours rather than hundreds.
A philosophical task manager, based on notable philosophers such as Baudrillard, Foucault, etc. It's been a fun game of reading these works, trying to distill their essence, and putting them through the LLMs to see what comes out. Currently it's not one task manager but rather many prototypes, and I'm trying to reconcile various design issues. In the same way that Marx's "philosophy" is essentially a "praxis" of political and economic power, which does not presuppose a "truth" but rather creates one, my task manager is a pragmatic tool for establishing the meaning of life.
My dotfiles.
Specifically, I'm working to stop working on them.
I'm working on an autonomous snow clearing robot: https://www.frost-e.com/
The basic idea is that instead of trying to solve the general case in robotics (ie, the humanoid companies - multiple tasks, multiple environments, generalization, etc) which needs billions of capital, can you efficiently deploy current tech to robustly do one task in one environment for cheap.
As a side project, I've been working on a RAG/MCP/agent service with workflows, for security: https://mappedsky.com/seizu/
This primarily builds on top of cartography (https://docs.cartography.dev/), which I'm a contributor to, which pulls infrastructure, SaaS, etc info into a Neo4j database, makes edges, and has a standardize ontology, so you can do attack path analysis, vulnerability assessment, inventory, etc.
Seizu uses the RAG built by cartography, and adds reporting, graph exploration, automations and more on-top of it. It currently ships with a couple pre-made automations:
* Automatic dependency vulnerability fixes, like dependabot, but uses codex/claude/opencode in a sandbox to do so, which ensures it doesn't just bump a version in the file, but actually assesses whether the changes in dependencies need code updates, and gives you a thorough overview on if changes aren't needed, and why changes were needed if so. If tests break, it also watches and fixes them automatically. * A workflow to run the cartography sync, making it possible to split apart the stages, run portions of the sync on different schedules, in parallel, etc, while also making it possible to properly trigger downstream syncs, or other types of workflows. * Generic AI agent workflow, which can be staged, run in parallel, or chained together via inputs/outputs, using Seizu's own custom agent (which has good LLM provider support, sandbox support with files, etc)
Workflows are built on-top of temporal, allowing it to be quite flexible.
wellbody.me - a simplified approach to body health. we condense a progression system into 3 daily actions to keep you focused on your goals.
Autonomous mathematics research. If you like reading articles about mathematics, you might enjoy this:
https://alethean.org