alt Hacker NewsAsk HN: What Are You Working On? (March 2026)
What are you working on? Any new ideas that you're thinking about?
Comments
Still on time. It is almost two months now and this is such a deep subject and there are so many little tricky bits that I wonder if I will be able to complete the thing but there is still (slow) progress. I never suspected the amount of hard work that goes into building something that is stable at the nano second scale. But I'm becoming more appreciative every day ;)
I'm currently working on https://FaunaMap.app (web/Android/iOS) which lets nature enthusiasts see which birds, mammals, reptiles, etc. can be expected at any location worldwide using the past 20 years of GBIF observation data. It features an interactive global sightings map, color-coded to quickly spot recent rarities nearby. Public birding hides from OSM and a hotspots heatmap are included for trip planning. Users can quickly log large numbers of observations in the field. Observations with images appear in an Instagram-like feed for interaction with other users. Personal species life lists (global, per country, and custom locations) are also included. Feedback is very welcome :)
Building grith — OS-level syscall interception for AI coding agents.
The problem: every agent (Cline, Aider, Codex, Claude Code) has unrestricted access to your filesystem, shell, and network. When they process untrusted content — a cloned repo, a dependency README — they’re prompt injection vectors with full machine access. No existing tool evaluates what the agent actually does at the syscall level.
grith wraps any CLI agent without modification. OS-level interception captures every file open, network call, and process spawn, then runs it through 17 independent security filters in parallel across three phases (~15ms total). Composite score routes each call: auto-allow, auto-deny, or queue for async review. Most will auto approve - which eliminates approval fatigue.
Also does per-session cost tracking and audit trails as a side effect of intercepting everything.
https://monohub.dev — a new EU-based (hosted and developed) GitHub alternative. Currently, it has a file browser and a PR review tool. Started off as a personal tool, but grew enough to consider offering as a service.
I posted about it recently on HN (https://news.ycombinator.com/item?id=47199062):
It is at a fairly early stage of development, so it's quite rough around the edges. It is developed and hosted in EU.
I have started developing it as a slim wrapper around Git to serve my own code, but it grew to such extent that I decided to give it a try and offer it as a service. It doesn't have much at the moment, but it already has basic pull requests. Accessibility is high priority.
It will be a paid service, (free for contributors) but since it's an early start, an "early adopter discount" is applied – 6 months for free. No card details required.
I would be happy if you give it a try and let me know what do you think, and perhaps share what you lack in existing solutions that you would like to see implemented here.
For over a year now, I’ve been working on Compute Prices (https://computeprices.com).
It’s been a great way for me to better understand the cloud GPU industry, learn about data collection, normalization and use agentic coding to build a side project.
One thing I’m working on is distinguishing spot vs on demand prices and listing those separately. Also, including inference pricing for non-text AI models.
What features or data would you like to see me add next?
I finally, after a couple of years off, picked up work on a Standard Ebooks[1] edition of The Book of the Thousand Nights and a Night. I made a new years’ promise to myself to get it as done as possible by the end of the year. If it doesn’t happen I’m not going to be too hard on myself though, given the 5K endnotes and 1.5M overall word count. But we’ll see!
I'm building AthenaOS: https://athena-os.ai/
Basically OpenClaw but with investing dashboards for my portfolio, additional tools specifically for investing, and exploring an AI-Human collaboration on researching economics (check the 'community' tab).
The data models are all in markdown and Excel so that there's no lockin and you can manually edit positions, personalities, etc.
This comes from frustration around most investing tools basically scraping your personal data + forcing you to lock into subscriptions. I think it's now possible to just vibe code most of what one needs, aside form raw data subscriptions.
It's all open source, too: https://github.com/wgryc/athena-os
An accessible color palette editor for creating branded palettes built from the ground up that pass WCAG/APCA contrast rules (which is much quicker and less of a headache compared to doing manual contrast checks and fixes later):
https://www.inclusivecolors.com/
The current web tool lets you export to CSS, Tailwind and Figma, and uses HSLuv for the color picker. HSL color pickers that most design tools like Figma use have the very counterintuitive property that the hue and saturation sliders will change the lightness of a color (which then impacts its WCAG contrast), which HSLuv fixes to make it much easier to find accessible color combinations.
I'm working on a Figma plugin version so you can preview colors directly on a Figma design as you make changes. It's tricky shrinking the UI to work inside a small plugin window!
https://telephone.health, which shows how well LLMs can take narrative medical text, convert it to a structured form (FHIR R4, for application consumption), and then convert it back to narrative text for human consumption.
Interesting findings include Mistral doing better than Gemini 3 Pro in certain usescases, cross-LLM works better than one LLM to another, oh and - the cost all of of this. So, so expensive.
I've been working on my local server mocking tool Decoy: https://decoy-app.com/
I've built it earlier and also did a Show HN, now I am going through some of the steps that get recommended to me such as creating Product Hunt launches, etc. But I am struggling a bit with the concept of PH. What is the audience? People into new apps? It all feels a bit desperate to be honest and this app is just a hobby side project, I am not.
So if anyone knows of a good way to get some attention to my useful fun tool, please let me know.
Still working on https://kavla.dev
I have worked with data for a while. I feel like our tools could be much better when it comes to "flow". I want an experience where you don't need to alt+tab to slack/images/another query. What if we put it all on a canvas? That's what Kavla is all about!
Since last month I've done a lot of improvements to the editor to make the "flow" better.
I've also read up on HMAC, Nonces and fun encryption stuff to create read only boards.
Here's one where I look at stack overflow survey for databases: https://app.kavla.dev/v/mqhg54o319doya4.67dbfee1ccd6caf638d3...
Snowflake users apparently make the most money!
Been working on https://localhero.ai, its my service to automate on-brand translations for product teams. I've been doing outreach to Swedish companies/people, getting some good interest from a few that want to automate their localization workflow but don't want the work of maintaining own solutions. Even though you can build a version working with coding agents these days, there is a lot of stuff around it to make it work well over time in a product org. On the tech side for Localhero, one thing I've been working on how it learns from manual edits. Like when a PM or designer tweaks copy in the Localhero UI, those things now better feed back into a translation memory and influence future translations. It's like a self-learning loop, turns out a pretty nice combo of using old-school techniques and offloading some work to LLMs.
Also been spending some time on my old side project https://infrabase.ai, an directory of AI infra related tools. Redesigned the landscape page (https://infrabase.ai/landscape), going through product submissions and content, optimizing a bit for seo/geo.
I'm making a PC game called Doggy Don't Care. You're a dog left at home alone getting up to mischief https://store.steampowered.com/app/2438180/Doggy_Dont_Care/
Still working on
- https://github.com/rumca-js/Internet-Places-Database - database of domains and youtube channels
- https://github.com/rumca-js/crawler-buddy - web crawling / web scraping tool
- https://github.com/rumca-js/webtoolkit - web crawling toolki
- https://github.com/rumca-js/Internet-feeds - feeds databse
- https://github.com/rumca-js/Django-link-archive - RSS reader
Agentic code construction has broken traditional models for code review - the volume is just too high for humans to keep up with.
There are some good tools out there for automating pr review; IMO, they don't catch enough, and they catch it too late.
I've been experimenting with some ideas about a very opinionated AI code reviewer, one that makes an ideal tradeoff between cost and immediacy (eg, how soon after composition does the code get feedback).
Currently in an invite-only alpha, but check out the landing page and lmk if you'd like to be a trial user!
I am building ReifyDB(reifydb.com), a database for live application state.
A lot of existing databases are storage first, with everything else built around them. I have been exploring what it looks like if the database is closer to the application runtime itself, where state is live, queryable, and easier to reason about directly.
One thing I am prototyping right now is database-native tests.
Basically: what if integration tests were a database primitive?
CREATE TEST test::insert { INSERT test::users [{ id: 99, name: "Ghost" }]; FROM test::users | FILTER id == 99 | ASSERT { name == "Ghost" }; };
So not a wrapper, not a framework, not an external test runner.
A real test object inside the database.
The idea is that you could run these before schema changes, and make stored procedures or other database logic much easier to test without leaving the database model.
Still early, but it feels like one of those things that should just exist, especially for databases built around live application state.
https://github.com/AzimovParviz/openblaster/tree/qt6
I wrote a CLI utility last year to control my SoundBlasterx G6 DAC (can only control LED colour and EQ bands) without needing to use Creative's windows only program (I am mostly a Mac + occasional Linux) user.
Recently downloaded Qwen3-coder-next 80b model and been vibing with it to introduce Qt6 and write a dead simple (aka ugly) crossplatform GUI to it so that other people can use it on their Macs and Linux machines. Letting a LLM wreak havoc on your project feels bad, I constantly have to reign it in and rollback the repo once it starts looping due to writing something that doesn't compile, making it going back and forth between doing and undoing changes.
Happy March everyone,
I am working on Entangle, https://entangle.cloud something for me to learn and play with LLMs and AI.
It is not ground breaking but let your website to have an AI chat bot / agent with minimal integration effort. Also was a good way for me to learn how to keep things safe, prevent prompt injection etc.
Looking for feedback and feel free to give it a try, happy to try it with your project documentation or developer docs.
I’m building a decentralized Drone-as-a-Service (DaaS) orchestration layer that treats aerial robotics as a simple API endpoint.
The system allows users to submit a JSON payload containing geocoordinates and mission requirements (e.g., capture_type: "4K_video" | "IR_photo"), the backend then handles the fleet logistics, selecting the optimal VTOL units from distributed sub-stations based on battery state-of-charge and proximity.
Two choropleth map projects I've wanted to make for a while:
https://housepricedashboard.co.uk - shows a visualisation of house prices in England and Wales since the 90s, with filters for house types, real vs nominal, and change views over time
https://councilatlas.co.uk - similar structure to the above, but focusing on local council datasets. The idea is to make it easier to compare your local council's performance against the rest of the country.
I have built npm for LLM models, which lets you install & run 10,000+ open sourced large language models within seconds. The idea is to make models installable like packages in your code:
llmpm install llama3
llmpm run llama3
You can also package large language models together with your code so projects can reproduce the same setup easily.
I've been working on two small projects recently.
1. Live Kaiwa — real-time Japanese conversation support
I live in a rural farming neighborhood in Japan. Day-to-day Japanese is fine for me, but neighborhood meetings were a completely different level. Fast speech, local dialect, references to people and events from decades ago. I'd leave feeling like I understood maybe 5% of what happened.
So I built a tool for myself to help follow those conversations.
Live Kaiwa transcribes Japanese speech in real time and gives English translations, summaries, and suggested responses while the conversation is happening.
Some technical details:
* Browser microphone streams audio via WebRTC to a server with Kotoba Whisper * Multi-pass transcription: quick first pass, then higher-accuracy re-transcription that replaces earlier text * Each batch of transcript is sent to an LLM that generates translations, summary bullets, and response suggestions * Everything is streamed back to the UI live * Session data stays entirely in the browser — nothing stored server-side
---
2. Cooperation Cube — a board game that rotates the playing field
Years ago I built a physical board game where players place sticks into a wooden cube to complete patterns on the faces.
The twist: the cube rotates 90° every round, so patterns you're building suddenly become part of someone else's board. It creates a mix of strategy, memory, and semi-cooperative play.
I recently built a digital version.
Game mechanics:
* 4 players drafting cards and placing colored sticks on cube faces * The cube rotates every 4 actions * Players must remember what exists on other faces * Cooperation cards allow two players to coordinate for shared bonuses * Game ends when someone runs out of short sticks
---
Both projects mostly started as things I wanted to exist for myself. Curious what people think.
* Reconstructing `$HOME/git` after an unfortunate `rclone sync` wiped it. It was not fun discovering that my backup was on the wrong `git` directory and also that I'd not committed some stuff for ... years. Lesson learned, etc. * Implementing three new bot ideas. * Trying desperately to stay out of the "we must vibecode!" juggernaut's path at work. * Wasting hours having to manually download Every. Single. Model. from my MyMiniFactory library because they don't provide an API, a bulk download, or a sync to something like Dropbox. (1500 down, about 4000 to go!)
My friends and I would often sit in pubs trying to guess the top 10 most streamed songs for a musician so I had a go at turning it into a browser game.
Had a lot of fun doing it and it's more user friendly than trying to read off lists from stream aggregators online that have a lot of ads etc.
First ever go at something like this so any feedback is appreciated!
Building wireless (LTE-based) sensors for most major horticultural sensing needs. Measurements include:
- CO2. Side note: I was surprised to find that most (all?) CO2 sensors used in closed plant production setups are not meant to operate below 400 ppm.
- Air temperature, pressure, relative humidity
- Photosynthetically Active Radiation (PAR)
- Addons like: wind speed, wind direction, soil moisture and Electrical Conductivity (EC)
- The coolest and most challenging: pH, EC, and flow rate
The hardest part has been running everything on battery while maintaining accuracy and using LTE (2–4G) and not common LPWAN options like LoRa. I'm primarily a software guy, so the learning curve has been huge.
I’m working on an R7R7-small scheme called Sable. The focus is on having good Windows support, vscode support, Lsp and Debug adapter protocol support. It is closer to SBCL and is image based, and builds with just the platforms native c compiler
I finally decided to try and make a note taking tool I've been wanting to use. https://chrononotes.com/
As many here, I've found that a single text file is all that I really need, but found that it makes it difficult to keep track of a variety of things. I was also trying to use the file as a simple project tracker, adding some tags like [BUG-N], and updating them by hand. Eventually, it became difficult to track the progress of things, since I had to jump around the file to look for updates.. or use grep.
I condensed the idea to just that - a very simple tool which manages "trackers", and has a simple filtering built in to "trace" the updates. I've been using it, since I've added the BE, and dogfooding it a bunch. Would love for fellow note takers to take a look. It's not perfect, but I'm keeping it around for myself :)
I've finally finished the long-abandoned project that I've been meaning to build for a while.
If you use Stripe Billing for subscriptions, your customers can specify reasons why they cancelled (e.g. too expensive, not using it, switched to competitor, etc.). However, to access those, you either have to use Stripe Sigma or pull them from the API. I wanted to build a more convenient way to access those (and also act upon them).
I've submitted the app to Stripe's App Marketplace, but I have a limited number of test invites to send out if you're interested (I will happily waive your subscription for 3 months).
Building a new kind of news site, featuring updates from primary sources.
We're constantly pulling info from official sources, and using AI to group and summarize into stories, and continue to share reporting from trusted, vetted journalists.
The result is news with the speed and breadth of getting updates straight from the source, and the perspective and context that reporting provides.
Still ramping up, but I'd love to hear feedback:
Very small app for framing photos on Mac/ipad/iphone. I know some web alternatives. But when i wanted to add borders to photos I took to upload to IG i used imagemagick. I want to make something stylish(no tacky borderds), also usable for batch processing. I have prorotype and need to fix some issues.
Here is my fun mini-project:
https://apps.apple.com/us/app/kaien/id6759458971
I wanted a way for my kid to learn the alphabet, but without a UI that looks & behaves like a slot machine. It's all maximally slow, relaxed and designed to be easy to put down.
Working on Fairvisor — a tool for API governance and usage control.
The problem: one buggy integration, scraper, or infinite retry loop can suddenly explode your API costs or overload infrastructure before anyone notices.
Fairvisor acts as a guardrail in front of your API:
per-tenant and per-route rate limits
request budgets and soft/hard caps
anomaly alerts for sudden spikes
The edge component is open source (OpenResty / Nginx + Lua) and the SaaS part provides policy management and audit.
Still early, validating whether teams would use something like this instead of building internal scripts.
Delving on to personalized software again now that AI agents code it for me with little hand holding.
Instead of building scrips here and there I'm attempting my own everything app. Lets see if that's a good idea :).
Starting with android home screen and widget sync to server/desktop. So e.g. calendar notifications happen desktop first, then escalate to mobile etc. Also phone as mousepad (for using my projector from bed). Just feature creeping it all in without regrets so far.
In general I feel like feature creep needs to be reevaluated. UX must not be destroyed, but features cost less now.
rainy-city.com! rainy-city.com is an ambient rain sound generator that is also a kind of city simulation. it is my recurse center project. it's suppose to be more of an ambient experience than a city simulator. it's a total work in progress, I've implemented buildings but haven't made a PR yet because they don't really work the way I want them to, and so I had to rebuild the tiling for them. So right now, there is no city. lol. just rain. but eventually it will have all this stuff you would expect. there are whales.
I've been working on an MQTT Broker/Topic tree explorer. It's intended to help someone understand what kinds of data are moving through their broker, what the busiest data is, and just generally be pretty. https://ryanbateman.github.io/mqtt_vis/
OSS, MIT licensed. Feedback welcomed!
I am building a command line package manager for AI models. It’ll make installing and running models locally incredibly easy.
Checkout: https://llmpm.co
I’m working on a small deployment tool called push2start[1]. The goal is to make Docker Compose apps easier to deploy from a laptop to a remote server without a lot of custom scripts. Right now, I’m trying to keep Compose as the source of truth and avoid creating another mini orchestrator. You point it at a compose file, it transfers only the images that need to come from local, then deploys and gives status, logs, and restart controls from one CLI.
I am building a better console.log, for humans and agents
I’m learning Godot and writing a basic game for my kids.
Coding agents are amazing and make me (feel) productive, but they really suck the fun out of programming.
I’m sure it’s possible to create a Godot-based game with an LLM, but I’m not sure how, so I’m forced to do everything the old-fashioned way – reading the docs.
Working on a software trial automation infra.
While working on another project, I needed a very simple service I could setup in a few clicks, which would take my docker compose and manage the spin up and tear down of ephemeral VM automatically when triggered by a signup on my landing page.
I couldn't find anything real simple, so I decided to build it. Currently working on it.
Any feedback will be much appreciated.
I made my own AI personal assistant:
https://github.com/skorokithakis/stavrobot
It's like OpenClaw but actually secure, without access to secrets, with scoped plugin permissions, isolation, etc. I love it, it's been extremely helpful, and pairs really well with a little hardware voice note device I made:
I'm building web-based CAD software for woodworkers. Not a plugin, I'm starting from scratch. I'm aiming for it to be intuitive for non-technical users (think SketchUp), while also offering some of the more powerful tools of "proper" CAD tailored for woodworking: simple parametric workflows, cutting layout optimization, built-in tools like chamfers and joints,...
I've been slowly hacking on game ideas on and off for the better part of a decade and I've finally switched tracks and trying to seriously build something full time
I've given myself 6 months
It's a bit scary basically 180ing like this but I figure if I don't try it now I never will
I've already started prototyping various ideas, and to be honest just sitting down and spending time doing this has been really quite lovely
One thing I'm finding fun is slowly unearthing what I actually find interesting
I started with messing around in minecraft and tinkering with rimworld-like game ideas, but I'm slowly moving away from them as I've been tinkering more and more
Don't get me wrong, I do want to revisit them at some point in the future, but I do find myself circling more around narrative, simulations and zachlikes
It's a bit of an odd mix and in some ways they look like paradox style games, but I'm well aware that taking one of those behemoths on is going to be a bit silly, so I'm trying to slim down until I get to a kernel that I actually find enjoyable tinkering with
A toy if you will
Currently I'm trying to work out if there's anything interesting in custom unit design, basically unpicking how games like rollercoaster tycoon's coaster design maps to stats like excitement ratings and seeing how that might mix with old school point buy systems
It feels like it might be small enough to be a good toy and I'm having fun tinkering with it, but I have no idea whether other people will xD
It might honestly be too niche for anyone and I've successfully optimised for an audience of one :shrug:
https://symgraph.ai/ - AI-Powered Reverse Engineering Inside Your Disassembler
Open-source plugins for Ghidra, Binary Ninja, and IDA Pro that bring LLM reasoning, autonomous agents, and semantic knowledge graphs directly into your analysis workflow.
Coming soon: A supporting online service. The VirusTotal for reverse engineering. A cloud-native symbol store and knowledge graph service designed for the reverse engineering community.
- Submit files for automated reverse engineering and analysis
- Query shared symbols, types, and semantic knowledge
- Accelerate analysis with community-contributed intelligence
- Versioned, deduplicated symbols with multi-contributor collaboration
Working on an all in one "platform" that runs multiple different checks on a website/domain. Got sick of having to run different checks in different places to ensure everything is at it should be - SSL, HTML, SEO, redirects etc.
Extended the checking to monitoring and change detection/alerting. You can try for free at https://www.augsentric.com - built for my own needs, but made it for others if there's interest... feedback welcome
Side project - plan mode and code review annotations for coding agents (ui that integrates via hooks): https://github.com/backnotprop/plannotator
Main gig: Trusted agents. We just shipped hardware based signing to web bot auth protocol.
Working on improving AI Nexus - my custom LLM frontend that exposes all AI Models in a single interface: https://getainexus.com/
Building a tool that automatically generates living infrastructure diagrams from your cloud anccount and turns them into real-time incident dashboards. Think Figma meets Datadog - beautiful visualization that updates during outages to show you exactly what's failing and how to fix it.
I wrote this Telegram bot that translates any video with AI-generated subtitles in about 2 minutes. You paste a YouTube, TikTok, or Instagram link, pick your language, and get back the video with burned-in subtitles.
It started because my wife watches Chinese dramas and new episodes never have subtitles for our language. Turns out thousands of people have the same problem — Arabic speakers watching anime, Russian speakers following Turkish series, Persian speakers catching up on K-dramas.
Supports 40+ languages, works with any video link or direct file upload. There's also a Mini App inside Telegram for a more visual experience.
I'm working on Firefly, a programming language for full stack webapps:
https://www.firefly-lang.org/