alt Hacker NewsAsk HN: What are you working on? (August 2026)
What are you working on? What have you been curious about lately?
Comments
Working on https://letterspractice.com
This is a high efficiency, narrowly-scoped, low screen-time early literacy app for families with kids aged 2+.
A graph-based code editor https://miralume.dev
I am working on another iOS to help me learn Japanese, this time for explaining grammar. Put in a Japanese sentence, tap Analyze and you get grammar detected, highlighted, explained. There is a public testflight if anyone is interested https://setsumeiapp.kulman.sk/
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.
I've been working on this for a while now. It creates a sort of AGENTS.md or CLAUDE.md table for your database which holds all the joins, definitions, formats, and relationships, as well as what are popular criteria sets and combinations.
And it populates it into a bi-directional visual (round-trip) SQL editor. Make changes in the UI or SQL text, it goes both ways.
It's a single HTML file.
I've been working on two projects.
## Tab Highway: Browser extension focusing on the UX/process of decluttering
It turns tab cleanup into a game.
When you pile up too many open tabs or whenever you want, it opens a night drive where each tab is a road sign you drive through or send to the trash.
[GitHub](https://github.com/pravj/TAB-HIGHWAY)
This is still a work in progress. It works on my machine, but not fully distribution-ready.
## A short satirical blog on the overly-expressive/interactive web experiments fueled by coding agents.
Ironically, the project listed above falls in the same category.
[Woodworking Has Been Achieved Internally](https://hackpravj.com/blog/woodworking-has-been-achieved-int...)
I resurrected an ancient UIUC x86 ASM game, built with some friends in the summer of 2002 for ECE291. Claude Code drove most of the work.
The first attempt was a full on x86asm->webasm transpiler which ran into some difficult stack issues.
The second attempt was successful via conversion from x86asm->c->emscripten->wasm [0]. It plays pretty well on mobile and just about any browser now.
Fair warning it is pretty difficult to win, unless you can figure out where the enemy weaknesses are [1].
[0] [code] https://github.com/gottebp/alan_parsons_project
[1] [play] https://particlefield.com/projects/alan-parsons/game.html
More things than I meant to, but one's live and three are landing soon. I quit someone else's work a little while back, spent some time building a game that I'll get back too to the extent I can support myself, and then started a things-I-want-and-others-might-want-too studio.
https://sponder.app (Live) - an RSS and podcast mutator. Filter feeds for what you want, merge, time-delay, and drip articles on a cadence (I like to start from feeds' beginnings). For podcasts, detect similar metadata to previous releases so you can skip reruns.
https://tellmeapenguin.com (Close! Come help test it) - Endless mad libs about whatever you want, and if it's a good story, you can keep writing more chapters. The kiddos love it, and you can share back and forth with grandparents and whatnot to keep sillily in touch.
https://nounogram.com/ (Web is live, apps soon) - Hybrid nonogram and crossword puzzles. Shade the grid to find the layout, then add the words where they fit. I built it to make the kiddo practice handwriting, and then it was more fun than I expected.
https://getyarnt.com (Free until I ship the Stripe integration) - A task tracker that's actively hostile towards you having tasks. Started as a joke, but I'm using it to track my own stuff and really enjoying it. If you have a manager who likes assigning you work, tell them to get Yarnt!
All of it's on https://hahamoment.dev/ in the likely event you don't need any of this but do want to turn on a light.
I was laid off in early June (was a staff software eng at Viam writing motion planning software for robots) and have been fiddling with electronics.
Most recently I made a portable AQI meter, since the air quality here in Central Oregon can quickly get bad.
It's a small 3d printed box containing a rpi pico, an 18650 battery, an e-ink display, and a SEN66 sensor.
I hit a button, it turns on, runs the sensor for ~40 seconds to get a good reading, displays on the e-ink, and then turns off. So there's zero power draw when it's off, and the battery lasts for ~1500 readings. Accuracy seems fine, the AQI, temp, and CO2 readings all match my permanently placed sensors inside and outdoors.
Other projects include using some of the old leftover robot scrap to automate irrigation in my apple orchard.
I've pivoted to game dev because I'm burned out on AI dev tools. I started working on a hexpunk skirmish roguelite deckbuilder (working name is Hexborne) where you place units and structures on a hex scenario map to build a base and eliminate your opponent. Imagine a combination of X-Com 2 and Starcraft using cards, set in a world reminiscent of Arcane but 10,000 years in their future when humans have ascended to godhood and colonized the multiverse.
It's been a lot of fun, Godot and agents have taken most of the pain I remembered from my childhood attempts at game dev out of the equation so I can focus on game mechanics, art and lore. I'm almost ready to start soliciting external playtesters (which I'll admit is a bit stressful), I just need to clean up the UI/card templating/art/tooltips/etc so the game is accessible out of the gate and I don't churn for reasons other than the gameplay not being tuned yet.
I've been working on https://smalldocs.org for some time. It as if "Claude Code & Microsoft Office had a baby". It's actually very useful if you develop with a terminal based coding agent and sometimes want to get out of the terminal view to dig deep into a topic.
I have had quite a bit of usage: https://smalldocs.org/analytics, which I'm very proud of, with many developers using it week after week.
It's free to use by the way.
Code is open source at: https://github.com/espressoplease/smalldocs
I'm working on a go (the game) server. still unpolished, but it's functional! currently hosted on a raspberry pi in my living room and proxied to the public Web via Tailscale. Rust backend, SQLite, preact frontend.
Working on an open-source alternative to Loom/Jam/Birdie/Screendesk
It captures the screen recording along with console logs, network requests, and session activity so you can get to the root cause faster
Six months ago I began writing and publishing one essay a month. Last month my wife and I hiked Fuego Volcano in Guatamela for our 20th anniversary. Here is my essay on that: A Country Within Myself - https://www.metanoia-research.com/dispatch-005-a-country-wit... .
I have been working on the marketing site for an ETL Product that I am ready to launch, and this thread made me realize that I should ask in another thread about recommended MoR/affiliate programs for current day.
This is meant to be something that will help business teams be better able to communicate what they want, while producing ready to deploy tech artifacts that do not rely on AI.
More importantly, I have been working on being a Dad - something we were told would be impossible naturally. Way more fun than being in front of the computer - our little one is nearly six months old now, and sleeping through the night until usually at least 8 AM, starting to roll over and enjoy tummy time more, and bringing smiles everywhere she goes.
I am working on a framework that allows creating very complex SaaS products with AI, that actually work. Very complex = +100k LOC of one-shotted AI-generated code. Codename o2p - outcome-to-proof.
It is a research-first method for designing or reviewing SaaS products. Its central question is: What customer outcome matters, and what evidence proves that the product delivers it?
Idea or existing product → Research real customers and current work → Identify their Jobs to Be Done → Design the minimum safe workflow, define the interface and system behavior (with a sophisticated scoring system) → Validate the workflow → Implement vertical slices → Observe proof of customer progress
I'm reviving my multi-player Snake game Snaketron. It's written in Rust and probably the most technically interesting thing about it is that it's built for auto scalable Cloud hosting. The websocket isn't tied to the host actually running the game. And it can scale and rebalance in the middle of the a match without disrupting the game.
https://github.com/Nan0pk/Rush-linux A Linux variant one dynamic power and peromance tuning daemon, because Why Not--- https://github.com/Nan0pk/local-ai-relay A relay to grab Ai web chat and run it as openai compatible provider to run batch jobs https://github.com/Nan0pk/Archiv An offline Ai first institutional memory harness, ingesting office docs and outputing cited formatted deliverables.
Ever since I read this https://paulbupejr.com/developing-the-optigap-sensor-system/ I've wanted to build a system that takes in a shape and adds bends and fiber routing so you can dynamically create readable objects. I was initially imagining wanting to add the bending points manually but I'm not so sure that's the best practice and I may want to think of a heuristic for good vs bad bend points before even considering the UI.
I also bought some cheap gloves from harbor freight and used a needle to route fiber through each finger to make a simple power glove like thing which I think I can connect to arm arm to make a cool robot game. I think of the two I am more excited for the first but its much more work than the second and my day job is getting more exciting than I would like so by the end of the month I'll have had to pick one to finish.
I am in the final stages of preparing to ship a 2.0 version of a fairly successful app, that has been shipping for a couple of years.
This is a complete, bottom-to-top rewrite (front- and backend). It's a huge improvement. I used an LLM, quite a bit. It's been a big help.
We're in the last stages of the first-wave testers. This is the one that affects the app UI. After this, we expand the test, and do bug fixes only.
Also, we need to start taking screenshots and app previews for the App Store. That requires setting up a test server, with anonymized (fake) users, so we don't compromise actual user anonymity. It's a bit of a pain, but it has to be done.
I'm working on a platform for context management that includes apps that operate on structured text formats (e.g. YAML, JSON, XML, MD) and provides editors for structured text formats (code editing using codemirror, markdown editing using prosemirror) and eventually all media assets (images, videos, audio).
The idea is you have a set of sources (e.g. Github, Google Docs, local file system), a bunch of transformations (templates, apps, llms/gen-ai, simulations) and sinks (communicated through changesets). The platform is like a cache layer that connects these together and allows for a graph of operations (source->executor->sink->source->executor->etc.) which allows for tracing the lineage of assets through a system.
It's an impossibly large vision made barely possible with agental coding assistance.
Work: Security engineering focused on securing authnz at hyperscale in our new agentic reality. What interests me? Agentic engineering, vulnerability discovery at massive scale. Vulnerability remediation at massive scale. Bounding agent non-determinism.
Personal: Local models. Agent harnesses. Track cars, turning wrenches.
Had agents build recently:
- Photo catalog app with all 100K photos in my library described by a local 9B parameter model (it is absolutely amazing and my favorite personal project in terms of genuien impact)
- Generic RAG layer (vector embedding + FTS in SQLite with RRF and document ingestiation layers for 1-2GB knowledge bases to power agent context for stuff I am interested in, neuroscience and philosophy is where I spend time on this right now, and basically every AI research paper publicly available to help agents tinker with local models). This is especially useful for local models, but it even helps frontier models because they can freely churn through without a bunch of web calls and such when you are on a topic thread or research.
- Qwen 3.6 35B A3B clean-up proxy, achieving a clean thinking + tool calling A3B that I can leave running for 24+ hours. This particular model gets stuck in long reasoning chains and needs encouragement to get on with it, especially on context tasks. It actually reasons well but then starts second guessing and loops. Built a deterministic framework to inspect its thinking and make it wrap its thinking up if it gets stuck and also to trigger calls to external models and inject solutions to problems when it gets stuck. The agent sees the solution injected from a remote model like it is its own thinking, kind of a subconscious it has no idea about. Metacognition in models, it is wild.
DSV4->GLM->Kimi is the "phone a friend" external brain pathing I built. As far as I know no one has shared all of the bugs I have 'fixed' in my project and gotten local Qwen A3B this reliable.
Hosting an XGBoost-model for MLB totals predictions. It runs every thirty minutes: incorporating weather at the hyperlocal level and joins with player attributes that are provably affected by that weather. It also provides advanced decision-making tools with a Kelly Criterion calculator built in. Model still needs some calibration for bet-sizing but is very close to production-ready.
Robust historical evaluations available at https://ballcast.app/validation
I got laid off on Friday so I'm using it as an opportunity to dogfood briansjobsearch.com again. I'm working on a Chrome extension that will help with autofilling job applications. It has dramatically sped up the time it takes to apply to each job even on complicated ATS's.
I have been working on automating many of the programs that federal prisons run to help newly released inmates with re-entry.
Example: https://atlanta-fci.bop.doj.dev/program/65545-inmate-account...
I’m building an image sharing site (just for me and a couple of friends) that tries to emulate the old days of Flickr and Instagram. Chronological feed, no recommendation algorithm, no reels, no ads, etc. Knowing that it’ll never be open to registration or “real” users takes all the pressure and expectation away and it’s just been fun to work on.
https://corvi.careers/reports/technology-stack-trends/
It tracks explicit mentions of ~ 181 languages, technologies, vendors, etc as seen in job postings. The motivation was that GitHub activity and developer surveys describe what people use or enjoy, but not necessarily what employers are requesting.
Been building a data ontology project i call Tirith - think palantir's gotham + their different modules on top. Can ingest from different sources (textual, structured or freeform for now, images + video coming maybe later), detect relationships between objects, view graph, geospatial view and rudimentary analysts view. Very raw but with the help of nlp and/or openai compatible cheap model like deepseek its already amazingly capable.
And second thing im building is out of sheer frustration as a father of 5 in the school info management - https://www.classtable.org - aims to be a full school management software which is easy to use for both school staff, students AND (personal itch) for parents to keep an eye on their childrens progress
I've finalized and have in beta testing an AI-based natural language search of properties from an MLS feed. On top of that, I'm continually tuning an AI agent that is able to do property research, comparisons, neighborhood and history analysis, etc.
[1]: https://mlsync.ai
I started making a programming language that is designed to encompass frontend, backend, database, container, and cluster, so that there it is not necessary to stitch together other languages.
It uses functional programming and event sourcing to try and effectively "solve" modern distributed web application development, so that your app code in this language can be extremely simple but not have bad architecture and gaps.
I'm probably oversimplifying the complexity, so I don't even know if it's feasible, but it's pretty fun to give Claude my crazy ideas and see what happens.
A music recommender website oriented towards albums.
Tell it your 5 favorite albums, get an immediate suggestions list. No sign up, free to use, integrated with Spotify to hear immediate samples (or the whole album).
Working on a web-based image editor with plenty of customisation and support for most Photoshop features.
Built overwhelmingly in vanilla JS with a couple of WASM modules for it.
It’s generally very performant across the app so far, but there’s still a handful of features that are functional but subpar in terms of performance that I need to address.
https://github.com/backnotprop/plannotator is still very useful for annotating html prototypes and markdown plans (as well as terminal agent messages). Was glad to see codex team pick up the feature requests in the Codex App.
About to release a fun paid version. All contributors get lifetime free access.
Better running agents run longer, they need better plans. (is my bet)
Finishing up my GSoC project with MalariaGEN. We're writing a new tool to estimate kinship coefficients between mosquitoes genomes and the project got a lot bigger than I initially planned haha. I'm going to tie up a smaller part for the GSoC submission in August and then keep working on the rest for a paper we're writing up. Also a big shout out to AVX-512 and sleef, for numerical work it is just chef's kiss. For anyone curious you can see the tools used to analyze DNA for disease vector control here, it's pretty interesting :)
I'm updating Metabase's Podman documentation to reflect using Quadlet instead of "the old way." It's my first open source contribution in quite awhile and while it's tiny I'm excited for it. It's a chance to get to know pi.dev, not to write, but to behave as a copy editor and fact checker. Since it uses Mistral on the back-end it's a slightly insane copy editor and fact checker, but that's almost better. It's forcing me to check everything I say.
I am working on Amber.jl an acausal analog electronics simulation tool driven by Julia code. Just like in PyTorch, the neural network is the code, here the circuit is the code as well.
Because SPICE wasn't built for programmers in mind. The code in 0.1.0 is almost ready and I'll put the github repo public soon.
I'm adding MoE support to the GPT-2 code from Sebastian Raschka's "Build a Large Language Model (From Scratch)". Just the minimal changes. Planning to do (or at least start!) a from-scratch base model training run on my own hardware before the end of the month.
I've always ran into an issue running arbitrary processes - whether it be for serious production work, or just toy scripts. So many choices with so many dependencies, featuring so many foot-guns. So I decided to take a stab at solving the problem. Would love any feedback or contributors :)
SystemG - An agent-friendly general process composer.
These are all really good comments and ideas! I created a tool to help facilitate, track and recover lost revenue from Stripe subscriptions for SaaS companies. Called RertyForge, www.retryforge.com It lets a failed payment be seen from a company owner’s perspective. Why things fail, what works best to recover it, what step of the process is most successful, who is most at risk right now. It’s bootstrapped and just getting it off the ground. Love any feedback or questions
I have two personal projects at the moment.
First, https://agency.nu/ a personal agent platform for people who aren’t necessarily developers, but still want to connect multiple Gmail, Calendar, and other accounts and automate workflows across them. Spent a lot of time making the interaction easy and natural, so it supports chat, realtime voice and telegram.
Second, https://vinarium.one/ something I primarily built for myself to keep track of the wines in my wine fridge, when to drink them, and what pairs well with a particular dish.
Both are very much scratching my own itch, but I’m curious if either resonates with anyone else here.
I love shared daily puzzle games. Wordle, crosswords, that kind of thing.
Swudoku is my take on a daily sudoku puzzle. All the numbers are on the board, you just need to rearrange them to complete the game.
Been running it for over a year now with daily plays from friends and family who enjoy it. Recently took a stab at an iPhone app for it as well.
I’ve been playing around with exploiting LLMs having ‘read’ every book ever written to make a chatbot to help people choose library books. You talk to it a bit about things you’ve read, and it uses the llm’s latent knowledge of books to suggest ones you might like.
It’s pretty fun! It might even work on your local library (I got carried away and spent a while making it work on a few thousand round the world - you can see the coverage here: https://leafle.nanosheep.net/coverage).
My own IDE!
https://github.com/paradise-runner/toast
It has all the parts I want (lsps, syntax highlighting, a file tree) without the parts I don’t (a bundled terminal, tracking, ai)
Since January I've been working on a storytelling system that fuses what I think is the best of interactive fiction / MUDs with the strengths of LLMs. The world model is entirely deterministic, but the LLM interprets the player and narrates the game and NPCs.
Also hacking on something similar to YC's QM but designed for small non-profits to ease as much back office work as possible.
I'm working on codecast. Something that started as a way to index / search and sync agent sessions from their jsonl files has turned into a high level desktop/mobile orchestration and product / steering layer above agent sessions. It's a huge productivity boon for me and for teams that have adopted it.
So I am working on building a low latency streaming dataflow framework - ala gnuradio but actually usable in production. The idea is to have deterministic latency as a first class objective and let all the other benefits fall out of that.
Additionally:
Live graph reconfiguration Timed events Easy simd support Support for non-mmu based architectures Imperitive configuration management and querying at the single runtime level. Can be compiled to wasm, and run in the browser (w/ coors headers). IO/user interface and any GUI is done via typescript.
I've been working on a simple multiplayer word game: http://alphabetsoup.club
It's a bit like NYT Spelling Bee, except it's multiplayer and only your best (longest) word counts toward your score. So, instead of trying to find as many words as possible, the game is more about finding one unusually good word.
Funny timing, I just made my first Show HN post for this a few minutes ago: https://news.ycombinator.com/item?id=49235347
Waiting for the remaining broadcast equipment for my LPFM (www.kpbj.fm) to arrive! Then we finally get on the broadcast band. We have over 80 shows at this point. We still need fundraising support. If you are in Los Angeles and want to support freeform community please reach out.
Had a problem this week helping a customer troubleshoot some SAML interactions. They couldn't install SAML Tracer in their environment and I got tired of digging through HAR files and manually decoding SAML assertions and the like so I made https://hartosaml.com/.
Just drop in your HAR file and get immediate insights into the SAML related events.
I’m building FitBee, a fast, ad-free, no-nonsense calorie and macro tracker for iPhone, iPad, and Apple Watch. The app is built with Swift and SwiftUI, with a Python backend. https://apps.apple.com/us/app/fitbee-calorie-macro-counter/i...