You select columns and then just drill down to create further joins. Change the SQL text and it updates the view.
I'm a CS undergrad would love feedback.
One of the most interesting applications for LLM's is writing SQL based on a schema, and I wonder if your tool could incorporate a "show me the books titles from authors who's name starts with T" and write that out.
Good luck!
Yes, I agree. Just as we need to check what LLMs produce when writing code, I think this could be a way to check what they produce when trying to write SQL.
As simple and personal as can be. Straight to inbox.
Optimized for coding agents DX through full customization and data appending by url parameters.
https://formvoice.com Appreciate any feedback!
I've been working on an open source LLM proxy that handles the boring stuff. Small SDK, call OpenAI or Anthropic from your frontend, proxy manages secrets/auth/limits/logs.
As far as I know, this is the first way to add LLM features without any backend code at all. Like what Stripe does for payments, Auth0 for auth, Firebase for databases.
It's TypeScript/Node.js with JWT auth with short-lived tokens (SDK auto-handles refresh) and rate limiting. Very limited features right now but we're actively adding more.
Currently adding bring-your-own-auth (Auth0, Clerk, Firebase, Supabase) to lock down the API even more.
https://news.ycombinator.com/item?id=44627910
In lieu of chatbots as the primary means of working with AI.
This is an approach that is human centered and intended to accommodate a wide array of possible use cases where human interaction/engagement is essential for getting work done.
https://demo.snapreceipts.fyi/
Mainly used by my friends right after we have a group lunch or dinner. You just upload a pic of the receipt after a meal and it parses out the items. We assign who got what and it calculates who owes what.
Makes the receipt splitting part super easy.
The goal is to have it behave like typescript for Go, where any Go program would compile out of the box, but then you can use the new syntax.
Featuring: built-in Set/Enum/Tuple/lambda/"error propagation operators"
It also have a working LSP server and generates a sourcemap, so when you get a runtime stacktrace, it gives you the original line in your .agl file as well as the one in the generated .go file.
I recently finish porting all my "advent of code 2024" in AGL -> https://github.com/alaingilbert/agl/tree/master/examples/adv...
When I wrote Go, I figured that I would eventually have to do something like that, to fix the glaring omissions in the language. And then I stopped writing Go, but glad to see that someone got around to it!
After that, I'm not sure. I have four big ideas:
1. (continuation) Another video, this one about my experiences writing a homebrew PSOne game
2. (useful) a command line tool (or native desktop app) that generates white noise
3. (fanciful) See if I can unpack FFVII's world map data into OBJ models and UV mapped textures. And then from there create a 3D world map in Threejs
4. (stretch) I would love an app where I could look out into the distance, and be informed what's on the horizon. Likewise ships in the sea / planes in the sky. I think it's doable with some OSM data, open APIs and a bit of high school math
Been fun to push Nanite and Lumen to the limit!
[0]: https://boris.kourtoukov.com/we-wade-awake-live-visual-perfo...
Earlier today I implemented "bbcodes" for bold, italic, underline, em (grey background color) and strikethrough. They way it works for bold is like this: b[text here]. If you want to apply multiple you can go bui[text here] for example, which would be bold, underline and italic text.
The most important features (for me) are: One Time Payment and 100% local and private. I don't send any data to any server. Just enough to verify license keys.
- Its one time payment, user can import any text, URL or ebooks and use the reader with read along text highlighting or export the audio as mp3 or m4a (audiobook specific format).
- Currently only supports MacOS with Apple Silicon I was doing Windows too but its making development slow, so I'm pausing that for now. - The most recent feature I added is Global Capture where user can setup some hotkeys to import any text and URL. Text parsing and extracting text is one of the hardest part of this. - Also, just added the a Reader view to website. Its goal is to mimic the app featuers as much as the browser limitations allow. I don't have a free Tier but a 7 days money back gurantee.
I mostly have a dev and engineering background but the most exciting aspect of this marketing and those stuff. Still trying to figure that.
I'd be happy to hear any feedback and ideas.
Edit: Only English at the moment. Adding more languages is in my plan but its very difficult for me since I don't know any other languages. But I think it would be great to add those as well.
I'm curious, what on-device text-to-speech engine did you use?
Well, kind of.
I've been working a ton on some variations and ports of it over the last couple months, but the problem is that I need funding.
So my plan is to setup github sponsors, where for each project people want me to work on, they can donate any amount, and for each $25, I'll work one hour on that project. It'll have a few related projects that all come from a unified vision I have for 90s.dev -- to be a full platform that recreates 90s-era development, from dos and qbasic, to win3 and vb3, not to mention assemblers for those who want it (see my show-hn about hram.dev).
The form stays online for 30 days. To keep the forms online for longer, I will be offering paid plans.
Our target platform is a 40 grams tinywhoop so it’s safe to fly everywhere and makes almost no noise :). A Roomba for mosquitoes!
The main plus compared to traditional systems is that a drone can cover an enormous surface in a short time compared to static systems or man-portable insecticide spraying. Our goal is to be competitive with ITNs against Malaria.
Some links :
https://hackaday.com/2025/03/25/supercon-2024-killing-mosqui...
I know of a Dutch company doing something similar. Focusses on pest detecting/mitigation in greenhouses atm: https://www.pats-drones.com/
It could be a great reconnaissance tool though.
Insect populations worldwide are experiencing significant declines in both abundance and diversity, with several studies reporting reductions ranging from 40% to 75% over recent decades. Estimates suggest that 5%–10% of all insect species have disappeared in the last 150 years, and some global meta-analyses indicate terrestrial insect populations are declining by close to 9% per decade.
> If you don’t want to kill flies, wasps, bees, or other useful pollinators while eradicating the tiny little bloodsuckers that are the drone’s target, you need to be able to not only locate bugs, but discriminate mosquitoes from the others.
> For this, he uses the micro-doppler signatures that the different wing beats of the various insects put out. Wasps have a very wide-band doppler echo – their relatively long and thin wings are moving slower at the roots than at the tips. Flies, on the other hand, have stubbier wings, and emit a tighter echo signal. The mosquito signal is even tighter.
Fascinating engineering! Doesn't seem like it would be possible but it apparently is. There's also more visuals at about 17 minutes in the video embedded in that article, the signatures seem fairly distinct.
My friend once came up with a joke idea for a solar powered ransomware drone that would fly to a random roof and jam wifi signal until someone paid it to leave.
Is the name a word play with "torgnole" at all, or does it mean something?
Most effort on https://wheretodrink.beer, collecting and cataloging craft beer venues from around the world. No ambition of being exhaustive, but aiming for a curated and substantial list. Since last month I've added a couple of minor things like maps and "where to go next" sections for each venue.
I'm debating whether or not I should add user accounts, and let people maintain venue bucket lists, venue endorsements. Also planning to reach out to the venues and ask if they agree to monthly or quarterly one-click information verification emails from us.
Other projects that receive less love are:
- https://drnk.beer, a small side project offering beer-related linkpages, and @handles for Bluesky (AT Protocol)
- https://misplacy.com, just a dumb and wrong AI landing page for now but was thinking to work towards a drop-in solution for SMBs around lost/found management.
- A platform for helping voluntary associations with repetitive administrative tasks (non-english so not linking. Trying to rank the pain points currently)
- A platform for structuring national soccer club history (initial brain dump idea phase)
- A platform for structuring writing prompts and collaborative fiction writing (initial brain dump / mockups)
For the next month or so I think I need to prioritize what to focus on after summer
Always interesting to see what others are building and doing. So thanks for sharing!
Also Plex for books (https://www.passagebooks.com/) but that has a much bigger scope.
Had a fun week fixing up the application so it’s 100x faster on 5 different axes, and it’s starting to feel really well polished. Also started to move from reagent to preact/signals in a long slow migration hopefully to hsx.
I also moved the critical algorithm logic into an independent Clojure file that is compiled (and tested) with cherry-cljs — I’m hoping to expand this to ClojErl and jank so I can have isomorphic Clojure code running on the browser, BEAM server, and native swift app :D
It’s getting really close to done, I’m using it now to study 18 different languages, including some really minor ones like Maltese, Welsh, and Cantonese (not sure if Cantonese is really a minor language, but definitely low learner resourced) and it’s easy, slick, and surprisingly effective!
Additionally, please try it out before assuming they’re just claims. I’ve been using it daily for 3 months in 18 languages (roughly one from each major language family), and been in pretty constant contact with native speakers.
True I cannot vouche for all 120 languages, and sure there is the occasional error in the lower resource languages. However, I have put in a lot of work to make sure I have a representivie sample, and the errors are currently well within an acceptable range — and I’m working hard to improve them!
Why would you need a UI if the basic way to learn a language is to speak to someone? (I suppose you meant graphical UI.)
Wouldn't a good STT/TTS interface be more appropriate?
Why have a visual interface at all? It’s more convenient to use, is way more engaging, and just better for learning. There is an audio component to the app, and I’m sure more and more audio components will be added, but I would be surprised if in-app audio ever exceeds half of my usage.
For example there is a shadowing exercise that is purely audio based (put in headphones and press play). But what if you want to see what a word means? Or see its gender/case/tense? Mark it as easy or hard, remembered or forgotten? I can look all this information up, plus read a few paragraphs of explanations in less time than I would take me to formulate a question.
Also, thinking about resurrecting http://opalang.org/. We'll see if I have the energy to work on that.
It's designed for sync, so rather than fetching you can hook it up to a sync engine (any!) to keep your front end in sync with your backend. It's built on Tanstack Query, making the sync engine optional, and a great path for incremental adoption.
The query engine uses a typescript implementation of differential dataflow to enable incremental computation of the live queries - they are very fast to update. This gives you sub ms fine grade reactivity of complex queries (think sql like joins, group by etc).
Having a lot of fun building it!
https://tanstack.com/db/latest https://github.com/TanStack/db
[0] https://apps.apple.com/us/app/reflect-track-anything/id64638...
Encryption uses Fernet (symmetric), and all decryption happens only at point of access. There's no data retention after viewing or expiration. Optional analytics give visibility without compromising identity. Users can get notified when their shared links was accessed by the recipient, and they can set passwords for enhanced security. Limitations include email-based signups and no end-to-end encryption (yet).
You can check it out at = https://www.closedlinks.com/
You can read the white paper here - https://www.closedlinks.com/white-paper/
The idea came to me when we were trying to find ways to manage Terraform secrets , CI vars were a no-go because people sometimes wish to deploy locally for testing stuff, and tools like Vault have honestly been a pain to manage, well, for us at least. So I have been building this tool where the variables are encrypted with `age`, have RBACs around it, and an entire development workflow (run ad-hoc commands, export, templating, etc) that can easily be integrated into any CI/CD alongside local development. We're using this and storing the encrypted secrets in Git now, so everything is version-controlled and can be found in a single place.
Do give it a try. I am open to any questions or suggestions! Interested to know what people think of this. Thanks!
[1]: https://kiln.sh
Appio lets you add mobile widgets and native push notifications to your web app within minutes, without building or maintaining mobile apps, hiring developers, or dealing with app stores. You can try it at: https://demo.appio.so/
If you’re building a web-based product without a mobile app, or just want to try Appio, I’d love to chat! You can reach me directly via https://my.appio.so/ or drop a comment here.
Would love feedback - in open alpha:
www.draftboard.com
GET /hello
|> jq: `{ world: ":)"}`
pipeline getPage =
|> jq: `{ sqlParams: [.params.id | tostring] }`
|> pg: `SELECT * FROM pages WHERE id = $1`
|> jq: `{ team: .data.rows[0] }`
GET /page/:id
|> pipeline: getPage
https://williamcotton.com/articles/introducing-web-pipe
I would love feedback!
The validation piece makes it feel a bit a bit like the Rails mindset for people who work better in FP.
I'd make a could of suggestions for the docs: Maybe a bit more discussion of how we'd test our webpipe code. I see why you've called them 'middlewares' but, maybe the term 'macros' or 'pipeline functions' might avoid confusion with express/connect middlewares
And thanks for the motivation to for figuring out a good way to talk about testing and generally clean up the (very messy) docs.
It’s comments like yours that give someone the drive to continue.
1. Software: An OS that masquerades as simple note taking software.
Goal is to put an end to all the disparate AI bullshit and apps owning our data.
I solved context switching for myself ages ago and now I'm just trying to productize it outside my 3 companies internal usage.
It also solves context switching for AI agents as a byproduct.
2. Ethics: Give Ai and proto-Agi a reason not to kill us all.
An extremely minimal, empirical naturalistic moral framework that is universally binding to all agents so AI won't kill us all. I view the alignment problem as a epistemic moral grounding issue and that the current pseudo utilitarianism isn't cutting it. Divine command, discourse ethics, utilitarianism, deontology they are all insufficient.
On parental leave with my third. We are on month 4 so I have (a bit more) free time in the late evenings after we put the older ones to bed.
It is possible to run Playwright inside a Chrome extension, however, it requires the Chrome DevTools Protocol (CDP) to automate a browser which really hurts the user experience, is very slow, and opens security vulnerabilities. Chrome extension APIs can accomplish maybe ~85% of the same functionality as CDP or Webdriver BiDi -- it isn't complete because of security features which shouldn't be bypassed anyhow. For example, instead of calling a function in a content script with 'script.callFunction' with Webdriver BiDi in Playwright, a function is called with chrome.scripting.executeScript(). It will be 2 or 3 more weeks before I post a PoC.
This is following my work using VSCode's core libraries in a Chrome extension exactly as they are used in an Electron app to drive VSCode and Cursor. The important part is VSCode's IPC / RPC which allows all the execution contexts and remote runtimes to communicate with each other. [0] This solves many problems I have had in the past automating browsers with a Chrome extension.
Atmos Sleep Lamp: A bedside lamp that reduces blue light at night and wakes you up more naturally with light in the morning [1]
[0] https://restfullighting.com/products/bedtime-bulb-v2-preorde...
[1] https://restfullighting.com/products/restful-atmos-preorder
Supports Postgres, MySQL, SQLite, MSSQL, ClickHouse. Includes AI export to generate DDL in any SQL dialect.
17.5k+ GitHub stars, Feedback welcome!
It's an AI video game sprite animator.
Supports deployments of your own apps as well as 15k+ other packages (postgres, airbyte, dagster, etc) via helm charts.
https://github.com/czhu12/canine https://canine.sh
Reason? Got sick of paying for the massive markups on PaaS but missed the simplicity and convenience.
https://www.youtube.com/watch?v=XlolXvBDmRY
Cubic chunks, full lighting engine, opting to be non-deterministic, everything is unsigned integer math except for rotation and rendering, multiplayer is mostly implemented, built to be able to handle heavy simulation.. the foundation work is almost done. Right now it's just a hobby to try to build the best thing I can build. I work on it because it's fun.
https://www.literally.dev/resources/marketing-to-developers-...
Now bootstrapping https://www.minute-master.com - AI formal minute generation for regulated firms primarily in financial services space but also free to use for charities.
It's kinda like Magic Wormhole without typing. It uses iroh for the p2p networking - on both ends, and also in the little web app that you use to scan the QR codes and start the transfer.
It's possible to install with nix and I'm working on other package managers. I'm targeting Linux and Mac.
It has a ticking sound, and the notifications remind you to stay hydrated, stretch and walk. I've used many different Pomodoro and I'm trying to consolidate the features I like the most from each.
Right now it works quite well on Linux and it should work on Mac.
Recently I also made a font for it! https://untested.sonnet.io/notes/433-how-to-make-a-font-that...
I'm also thinking about organising the usage patters, because over the past few years I've collected a few interesting groups: mental health focussed users, script writers, neurospicy folks, bloggers, squirrel enthusiasts. I'm thinking about this here: https://untested.sonnet.io/notes/how-people-use-enso/
Since we added MCP and the use of structured output to "spill" multiple return values into adjecent cells, it is the quickest way I know of to monitor competitors blogs everyday before my 09:00 meeting. And also the quickest way I know of to test new AI models. I have a sheet with SimpleQA, MMLUPro, or GPQA Diamond and testing a new model is a matter of adding a new column. The whole idea is to enable normal people (like, non-techies) to automate manual, repetitive tasks with AI like programmers routinely do.
You can at least play some games now though:
https://housepriceguess.com/roundup/v/holiday-destinations/p...
Enjoy. And yes that really was my wife playing one of the games for the first time in the video ;)
Last month I decided to take a subscription of my own for Claude Code to use in my personal time mostly for practice and educational purposes.
So the past few weekends and the occasional week night I've been vibe-coding a game for iOS/MacOS using Swift and SpriteKit.
I have some experience with Swift previously but not at work, so it's extremely experimental for me. However it's been going pretty well. Most of the hang-ups are Xcode configuration issues.
It's interesting to poke Claude a bit and discover what it's actually decent at and awful at.
Gameplay mechanics-wise it's been able to implement things as requested generally without problems.
UI elements like menu screens and such it has been almost completely unable to do no matter what prompt I give to it.
It's safe to say I would never call the codebase professional quality. However, the base game has been implemented well enough to play without bugs and I've been solidly impressed.
The other issue I've had is if I want to change project/target/build settings, Xcode doesn't provide an easy way to do so. You need to poke around the UI to find where these settings and file relationships are set and change them that way.
There's a project file that I believe contains them all but it's not intuitive to modify by hand.
I'll still need to implement some kind of "AI" opponent or hack together P2P networking to demo it though; playing against yourself is fine for testing, but not really how the game is meant to be played.
It's hand coded so far, but I'm hoping AI can be a big lift for churning out the multiple thousands of named special rules, as most of these are very simple (+1 here, reroll there, etc).
Any WH40k players out there? Love to hear your thoughts!
This is the first time I've ever actually released something with a monetization option, so I'll be interested to see where it goes. It's a small enough niche that I think I have several features that genuinely don't exist anyplace else, like the ability to lemmatize even heavily inflected words (a very common stumbling block for learners of Finnish).
A web app would obviously be much easier to monetize, but then I would lose the buttery smooth feel of the search at it currently exists.
Tsemppiä! It's not live yet, but when it is it will be at https://taskusanakirja.com/.
It makes answering customer emails 10x easier.
The magic are training templates which are templates that get suggested (and eventually auto-selected) and personalized by LLM for every reply.
Every reply sent trains it to auto select that training template for future similar customer emails.
The stack is Ruby on Rails and Postgres hosted on DigitalOcean. The LLM currently is Kimi K2 hosted on Groq.
For now, it only has a daily newsletter fully compiled by AI agents without any human intervention. I plan to add public listed companies (semiconductor, energy provider, etc) onto the platform. Already found lots of good data points that can be used by analysts, researchers or observers.
https://www.inclusivecolors.com/
Example with colors from HN to play with (the grays used for links and main body background, orange from the navbar, green from newbie usernames):
https://www.inclusivecolors.com/?style_dictionary=eyJjb2xvci...
The main features are it shows if your colors meet WCAG accessible contrast on a live UI mockup, you get quick and precise control over every color grade in a swatch (via editing HSL curves) instead of these being auto/AI generated, and it helps you create a full palette of tints/shades for each color rather just a handful of colors.
The idea here is to design your tints/shades upfront with accessible contrast in mind so you don't run into problems later. Most brand style guides I see only have around 5 brand colors, and when you need more tints/shades later to implement actual UIs and landing pages, you get into a conflict where you can't find contrasting colors to introduce that match the brand.
I've had interesting feedback about different workflows designers have so far. It's tricky to make a single tool that fits everyones workflow but I might end up with multiple modes e.g. easy but more opinionated, and more freeform but for advanced users.
I admit it has a learning curve at the moment but I'm not sure how simple you can make it without giving up control. I think once you get it though, you'll realise it's much easier to make a custom accessible palette than you thought.
You can use the CSS export in regular CSS projects directly e.g. via `color: var(--red-900)`, or something like `--bs-danger: var(--red-500)` for Bootstrap projects with semantic naming. The same export format works for Tailwind too because since version 4, Tailwind is mostly configured via CSS variables now.
I probably need to make this more obvious, but if all your swatches have the linked/shared lightness option set, you can pick lightnesses where all grade 500 colors contrast against all grade 100 colors, all grade 600 colors contrast against all grade 200 colors etc. so when you're picking colors in CSS, you know by design which colors will contrast without having to go check them.
Thanks, feel free to message me if you want some tips!
Most accessibility/WCAG guides say things like "Tip #1: Choose accessible colors", but they don't go into any detail about how you pick sets of colors that contrast, like text/background/border colors for buttons on different backgrounds, as if it's trivial. It's actually really tricky and can feel impossible in some scenarios until you internalize the basic rules and constraints.
I usually see people saying the opposite, that it's easy to pick accessible colors, when it's often not, especially when you have existing branding to stick to.
I am using cerebras for book translations and verb extraction and all LLM related tasks. For TTS I am using cartesia. I have played around with Elevenlabs and they have slightly natural sounding TTS but their pricing is too steep for this project. Books would cost a couple of hundred euros to process.
Still in beta and learning a lot from each customer we onboard. We're actually going through our own SOC2 assessment in August, which has been... educational. Recently added business continuity and incident tracking features. Trying to build something that's actually helpful rather than just another compliance checkbox tool.
If anyone's interested: humadroid.io or feel free to join our beta waitlist at https://humadroid.io/join-the-humadroid-beta-waitlist/
If anyone's been through the compliance journey, would love to hear what worked (or didn't work) for you!
Idea is to add a lot more NSFW stuff like sexy avatars and mocap animations, cinematic controls, even a marketplace of content and assets.
(Built for fun as I optimized my daily spending to get a year's worth of flights for free and friends wanted it haha)
First game in progress https://reprobate.site
Any feedback is welcome!
https://github.com/bugthesystem/Flux
Flux is a high-performance message transport library for Rust that implements patterns inspired by LMAX Disruptor and Aeron. It provides lock-free inter-process communication (IPC), UDP transport, and reliable UDP with optimized memory management for applications with low latency requirements.
This week I’ve been working on predicting upcoming paychecks with Nodejs so we can automatically decide how much funds to move into your budgets when you get paid. I pull the past 3 months of transaction data from our Postgres database using Prisma and run some analysis.
People think syncing and delayed transaction data is normal, and I’m working on changing that by having the budgeting built in to the checking account. Along with a high yield savings account, goal envelopes, bill envelopes, etc, joint accounts, etc.
The app is designed for older adults who enjoy reminiscing but struggle to organize their thoughts into a coherent narrative. The goal is to preserve their hard-won insights and pass them down—to family members who may be too busy to ask the right questions now, and to future generations who would otherwise never hear these stories.
I have a working prototype that allows me to test the interview flow, and I’ll soon be sharing it with friends and family for initial feedback. I’m now looking for a designer to collaborate on the next phase.
Design will be a critical part of this app. The way stories are visually presented will be central to the user experience and will likely determine the app’s success. If you’re a designer interested in this kind of work, I’d love to hear from you. Given the text-heavy nature of the app, experience with typography and content-focused design will be especially valuable.
I'm intrigued by the use of level set domains here. I've only encountered those in other type of numerical simulation where the intent is in avoiding surface meshing.
I suppose moving an object in this context is as simple as composing its level set function with a translation and rotation. However, deforming is non trivial, especially local deformations, right?
How do you efficiently resolve collisions? At the scale of an element, it seems to be a simple check of nowhere should both level set functions be negative. But how do you select the elements to check? Do you somehow keep track of only the elements traversed by the objects in a time step, or some other method? I would guess your method should be more efficient than intersecting meshes, is that what you've found?
I'm particularly interested by your mention of high-order boundary parameterizations, what do you mean by that exactly?
Sorry to bombard you with questions, I was intrigued by a combination of things I'd never seen together before!
Early Access for a new terminal emulator [0] bringing dead text to life. It's my professional dream to evolve our conception of terminals without bringing in the bloat of, say, electron (read: staying native).
>Do you have any new ideas you're thinking about?
I like the thought of dropping you into the terminal right on the browser. It wouldn't be the real thing, but having a toy to play with is superior to dry docs.
Everyone knows reading is the best way to build vocabulary, but many avoid it and turn to flashcards or spaced repetition because long texts can feel overwhelming, and they often have to refer to a dictionary.
This app gives users short, engaging passages focused on comprehension. While reading, users guess word meanings from context and find out whether they got it right by answering a few questions below. I believe this will be helpful for people who haven’t had much success with popular vocabulary learning methods.
I shared it on HN earlier (https://news.ycombinator.com/item?id=44543063), but it didn’t get much attention. If you're interested in novel learning methods or vocabulary, I’d love your feedback.
P.S. Login is required since the app uses LLMs to generate interesting passages. You can register with any non-existent email if privacy is a concern.
Part of another odd project, and testing how long the material holds up. =3
The game is mostly done, so I'm now focused on tooling to make it easier for me to craft each week's puzzle. I'm solving some interesting graph and optimizations problems
After working on and using many MCP servers, I hit couple of issues multiple times:
* Do I configure 2 MCP servers of same type for 2 different API Keys or do I manually update configurations all the time? (e.g. production and development environments)
* when I have too many tools enabled, I noticed that either I am hitting context limit too quickly or LLM is hallucinating when choosing a right tool
* Some MCP servers expose a lot of tools, I want to disable some of them forever, instead of doing configuration per AI assistant (first for Claude, then Cursor and so on)
* Most MCP servers are hosted by third parties, as a privacy conscious person, I do not want to share my credentials with third parties.
And I am building Aiko - AI tools marketplace: https://getaiko.app
NOTE: Gmail and Calendar apps are currently under CASA Tier 2 security assessment, hence not published to production. But you can see demo usage here: https://www.youtube.com/watch?v=ZgEy6Y1kfn4
Firstly a DevManual - for “any” software team/IT dept - how to think about the philosophy, history and practise of basically everything - release management, backup and recovery or IAM and security and marketing-by-engineering or CSS
It’s kind of “this much I know” and a working docker based OSS “software team in a box”
And the second one is really expanding on the philosophy - how software is changing companies and how democracy works with software
https://github.com/turbolytics/sql-flow
Building a company around a tool is hard. There's been some interest but streaming is kind of commoditized.
I'm taking everything I learned building it and working on a customer-facing security product, more to come on that :)
I also used the tool to generate an Adult Chess improvers FIDE rank list for all federations around the world. Here are the July 2025 rankings though it still needs major improvements in filtering - https://chess-ranking.pages.dev
------------------
Another idea that I have been working on for sometime is connecting my Gmail which is a source of truth for all financial, travel, personal related stuff to a LLM that can do isolated code execution to generate beautiful infographics, charts, etc. on my travels, spending patterns. The idea is to do local processing on my emails while generating the actual queries blindly using a powerful remote LLM by only providing a schema and an emails 'fingerprint' kind of file that gives the LLM a sense of what country, region, interests we might be talking about without actually transmitting personal data. The level of privacy of the 'fingerprint' vs the quality of queries generated is something I have been very confused with.
- Live PDF preview
- 100% client-side
- No sign-up required
- Includes a Stripe-style invoice template
- Built with modern web tech – simple to self-host or fork
Repo: https://github.com/VladSez/easy-invoice-pdf
Demo: https://easyinvoicepdf.com
Would love feedback, contributions, or ideas for other templates/features!
https://github.com/scallyw4g/bonsai
I also wrote a metaprogramming language which generates a lot of the editor UI for the engine. It's a bespoke C parser that supports a small subset of C++, which is exposed to the user through a 'scripting-like' language you embed directly in your source files. I wrote it as a replacement for C++ templates and in my completely unbiased opinion it is WAY better.
I'm 97% certain this is because the faster code leads to more page thrashing in the mmap-based index readers. I'm gonna have to implement my own buffer pool and manage my reads directly like that vexatious paper[1] said all along.
[1] https://db.cs.cmu.edu/papers/2022/cidr2022-p13-crotty.pdf
You make it sound like I was trying to troll everyone when we wrote that paper. We were warning you.
It’s been a journey but getting close to launching our first version to pilot customers in August. We use an enormous amounts of AI tokens every month to extract data not possible with any traditional player in this media monitoring space. Benchmarking competitors, tracking impactful discussions, and receiving actionable brand insights.
If you are currently using one of the big media monitoring companies, I’d love to chat!
https://github.com/rush86999/atom
Check it out.
Soon approaching a 1.0 release for sanctum once I get my brain out of vacation mode and into hacking mode again. A lot has happened this year and I am excited.
I will be talking about how sanctum and its cathedrals work at sec-t 2025 [2] so in full swing working on the demos and presentation.
check it out: https://mixpeek.com
Most recently, I've been incorporating a lot of improved UX design. The app has always used a playlist metaphor, i.e. your database of flashcards is your library and you can sort them in different ways and then hit Play to start reviewing. Within the review session itself, you go through the cards in the playlist in small batches so that it's less overwhelming, among other reasons. After every batch, the app returns to an overview screen so you can see what you've just reviewed so far in the session.
The challenge has been designing this overview screen so it's clear where you are in the playlist without making it overwhelming. I finally came up with a good design this week, which I was quite happy with: https://mastodon.social/@allenu/114921335089371494
I've been pleasantly surprised at how much of an improvement this new UI has made on how the product feels. The old UI only showed you the history of cards you've reviewed in the session and highlighted the most recent batch of cards. This new one shows you the full playlist, but redacts the contents of the playlist ahead, so you immediately get a sense of how much left there is to do, but without being shown what is in the contents of the cards. Interestingly, this has the effect of making you want to see what is in those cards, i.e. to keep reviewing!
It calculates optimal ways to load boxes into trucks or containers, considering stacking rules, fragility, and real-world constraints. You can drag boxes like 3D Tetris or upload photos to auto-estimate item dimensions. Recently added: batch-wise guided loading for warehouse use cases.
We're at ~$400 MRR and just opened up a 14-day free trial. Feedback, trials, and intros to logistics folks welcome.
Would love feedback, contributions, or ideas for other templates/features!
Absorbing low (male voice; 80Hz - 300Hz, not including overtones) frequencies normally takes a fair bit of dampening material, unless something like a Helmholtz resonator [1] is used. The paper shows that a ~100x100x12mm 3D printed Helmholtz resonator may entirely absorb 125.8Hz (in an extremely narrow band). I'm uncertain about transmission losses (i.e. volume of the frequency perceived behind the material).
So far, I have created/vibe-coded a script to take the inputs: frequency and tile dimension (it's square). The output is a 3D object (.stl) which can be printed.
Today I tested my 3D model, which roughly resembles the model in the paper (1mm roof & floor as opposed to 0.2mm, because of printing difficulties), by using a DIY'D impedance tube and publicly available software [2]. The print was meant to be tuned at 125Hz, but results showed 131Hz and absorption factor of ~0.42 (lower number as opposed to 1.0 may be due to inexperience with all of this; it may be due to an imperfect test setup).
My impedance tube is made from 96mm (inner) diameter PVC tube, a Visaton KT 100 V 4 Ohms speaker, an amplifier, Motu M2 audio interface, 2 Behringer ECM8000 measurement mics and some 3D printed adapters (to hold the speaker and sample).
Nothing to concretely publicise or share so far, but am thoroughly enjoying the process of digging into a field (acoustics) completely new to me, solely out necessity and/or frustration in the workplace.
Should anyone be interested, I will share my project with HN once it has progressed to where I have something written up or worth sharing.
[0] http://dx.doi.org/10.1063/1.4941338
Unlike its competitors, it uses proven research and techniques to measure the issues, as well as the improvements.
https://groundme.app/what-is-ground-me
Test users and early adapters are very welcome
Programming for me has become a lot more fun because of Claude Code. I get to spend more time planning and researching.
I have been working on https://codient.dev to be able to run Claude Code agents in the background without setting up a local IDE!
Animal colony management is largely managed in Excel sheets, with no integrations to related systems or hardware sensing. We're working on the spreadsheet problem first, so that biomedical researchers can share information about their colonies with other researchers at their institutions, and explore the lines that other labs have. This opens up collaboration options and makes it much easier for the research community to find out what mouse lines other labs have (and may borrow for their own experiments).
Currently in closed beta at Harvard
It's a fun project, because I have to do hardware for it, and that's outside of my current skill set.
Thanks to Claude, it works way better than I should've managed solo: https://www.procuratorai.com
Free signup to test: https://my.procuratorai.com/login (no help/intro yet, and I'm paying for tokens so not advertising it widely...)
Homepage is basically a one-shot Claude build using Nunjucks on Netlify (first time with both).
(Subscribe button is broken - still working on that...)
It started off because I wanted to see if I could print QR codes on a piece of paper and use these to detect people crossing a lap if finish line.
That proved more difficult than I thought, the QR codes were not easy to scan from far away and while moving. It is still in alpha stage.
What does work is a simple manual mode for people to use at their races.
Can be described as Astroneer-like setting, Teardown voxel physics, in a Valheim-like online multiplayer survival game.
Game isn't really announced yet but I've shown some videos of the tech: https://x.com/Alientrap/status/1909316208563732866 (On Youtube: https://www.youtube.com/watch?v=ZWISaUmvit4 ) https://x.com/Alientrap/status/1918024969939808654
Given the online/multiplayer aspect how difficult has the network portion been?
The C/C++ library can be easily embedded in host applications or plugins. It even runs on embedded devices, such as the ESP32. In addition, the project contains a Pure Data external and SuperCollider extension. There is also a third-party Max/MSP external: https://github.com/ddgg-el/aoo-for-max
For more background information, check out this article: https://www.soundingfuture.com/en/article/aoo-low-latency-pe... https://aoo.iem.sh/
The project is still in beta stage, but I hope to make a final release this summer.
Tomorrow, I'll start a brand new project, also related to the real estate industry and society.
Now connecting all the dots and building a backend for mobile apps. It’s already live https://calljmp.com
Fully powered by Cloudflare, unbeatable pricing, rls and app attestation, raw SQLite queries, and tons more.
Looking for early feedback and adaptors.
Intro video: https://www.youtube.com/watch?v=X_eKc6c5tDw
Why? SAP holds the most important data for companies that use it, but it's notoriously difficult to replicate this data consistently into a data analytics platform (think Snowflake, Redshift, etc...).
Couple of companies specialize in the SAP replication, but it's hard to validate the correctness of the replicated data, because:
- the SAP data is changing continuously and rapidly
- there are hundreds of tables and TBs of data
Usually it's the consumers of data downstream who notice that the data just "doesn't feel right".
Tracelake adds a validation layer on top of the SAP to X replication, which periodically compares the data between source and target and informs you about any missing / incorrect data, so you can tackle data quality issues proactively.
I'm starting with React Native to see if anyone actually ends up using it, and will go from there
[1] https://newspeaklanguage.org [2] https://multisynq.io https://chalculator.com/primordialsoup.html?snapshot=Amplefo...
[0] https://blog.walledgarden.ai/2025-05-27/wabbit-s2-mcp-openai...
The guiding principles are to create a fun, positive, safe space for kids and families to socialize and interact as well as empower kids to explore and understand technology as a creative tool and not just as something to consume content.
dedicated built for ai agents first, humans are welcome, too
Since WGU just started doing masters degrees in CS, I decided it would be a way to kill large amounts of time while getting at least a little out of it, so I registered for it.
I've been a professional software person for like fourteen years, so I was able to knock it out extremely quickly due to WGU's competency based stuff, so now I finally am able to put "MS" after my name.
I'm actually planning on doing a second masters from a slightly more prestigious university with a more theory-heavy degree [1], but it's nice to at least have an official graduate degree now. Hopefully it helps me find work a bit quicker, and if nothing else it's just kind of fun to pile up degrees.
And I've been vibe coding some maths educational tools and games for (my) 6yo: https://rupertlinacre.com/
We saw an increase in demand for individuals willing to build their own HackerNews, Product Hunt, or Reddit-like Community.
So, we built a SaaS platform for them, where they can launch their own community with their custom domain in just a few seconds.
Demo community - https://kocial.co
Get your own community here - https://kocial.net
This is a wheel I see people reinventing all the time, often for use in SaaS applications. The implementations are often underwhelming: function support is limited, documentation is sparse to non-existent and errors are typically only communicated at runtime -- if at all. Formula editors usually lack autocomplete, making them frustrating to use.
I've spent years solving all these problems (with a statically-typed language), and I'd love for others to benefit from the work. I have extracted the formula engine from our app compiler, so the library is nearly complete. The runtime part (evaluating formulas) has been rewritten in TypeScript. Next, I'll build a service around it to validate, compile and evaluate formulas -- which should be fun.
I'm planning to do a Show HN once I have a preview up and running.
So far I've cataloged about 1500 advertisements out of the ~100,000 in my possession. Of which that is probably only 0.1% of all the major material out there. It's going to take a long time! I'm going at a rate of about 10,000/year. I'm going to have to speed this up :) But I've gotten the process to catalog a full magazine down from a week to a few hours.
I'm thinking of ways to support the archive. I am doing original art from the ads I may sell, or sell really nice copies of rare ads.
So I will release my new data grid component based on my own toolkit, and if people want tweaks or "add these features", I will demonstrate them the toolkit.
The model is mathematically proven to converge to π, symbolizing natural harmony. So people can choose it not as a dream, but as a rational system for real well-being.
GitHub: https://github.com/contribution-protocol/contribution-protoc...
Nothing much to show other than one client, but I'm on the cusp of charging them monthly vs getting paid by the hour.
This is one of the most important performance features in a JS engine, as without shapes property lookups would be terribly slow. I'm looking forward to getting this working.
Easiest way to explain it is something like D&DBeyond but for indie games.
Link if anyone is interested: https://raze.cloud/
Getting together a very simple server (PHP), and very simple clients (UIKit and SwiftUI), and will publish a blog series on it (sort of like this[0]), once I get more used to it.
I need to really get comfortable with it before I do that, though.
It uses whisper.cpp under the hood and should be accelerated on most devices using the Vulkan backend
Not planning to do a lot during the promised Berlin summer though
Most people I know are using group texts for this, but I find that unsatisfying because my wife and I want to share stuff with ~20 people, but we don't want to be blasting all of them with texts all the time, or put those 20 people in a group text with each other. We wanted something pub/sub, but with the privacy of E2EE chat apps, and so easy to use our parents will use it.
It's a React app running on Cloudflare Workers, and there's an iOS app in the works using Capacitor; the E2EE is built on OPAQUE. There's a landing page/signup at freefollow.org if you'd like to learn more. I'm working on some demo videos.
https://github.com/search?q=repo%3AWebAssembly%2Fwasi-libc+s...
This is something I've been kicking tires on since my time at $BIGCORP; JSON without the bloat, Protobufs without the ceremony. I've drawn a lot of inspiration from MsgPack, CBOR, and Ion 1.1. Big emphasis on a tight set of core primitives, low-cost extensions, storing reused values/schemas, optional pre-negotiation, etc. That said, I've now been spending time trying to study the performance angle to make sure the design doesn't have a negative impact on encoding/decoding performance before committing to the implementation.
Regrettably nothing much to show (at least yet), but hopefully if nothing else it will become my go-to format for other personal projects that I work on.
After tinkering with game technology for years now, I'm pleased I've finally managed to use all that knowledge to create something (soon to be) releasable
https://store.steampowered.com/app/3407760/The_Night_of_the_...
With a strong rating weight system that can avoid (some) of the pitfalls of community ratings.
Right now videos must be added to be searchable, to comply with YouTube API rules. I'd hope that over time, with enough usage, the repository could contain many categories of highly curated content. (eg. Documentaries) that someone could find without having to browse various communities and opinions to get lists.
- Implementing a Convolutional Neural Network in pure Python to learn how it works.
- A Open AI Whisper to an embedding model pipeline to transcribe and summarize podcasts.
Only put the last three games of this past season so far, but I will probably add more each day (re-runs are still on until the new season starts in September).
https://www.virtualhospitalsafrica.org/
While medical records systems in much of the developed world are still shared via fax, we think there's an opportunity to leapfrog existing systems and have a cloud-based source of truth.
It's a tool to help teachers detect student assignments that have been written by AI. Unlike other solutions out there, it's an entire web-based text editor that analyses not just the final assignment, but all the keystrokes used during the writing process.
My theory is that analysing the final text only is a futile struggle - billions are being pumped into making LLM text look more human, so trying to make an assessment off final text alone is guess work at best.
I'm curious what folks think! Especially teachers, devs, and anyone navigating this space...
Not that your software is going to be useless. But as long as there is an incentive to cheat, new and better tools that facilitate cheating will crop up. Something else should change.
For both a keystroke based AI detector, and software designed to mimic human keystroke patterns, performance will be determined by the size of the dataset they have of genuine human keystroke patterns. The detector has an inherent leg-up in this, because it's constantly collecting more data through the use of the tool, whereas the mimic software doesn't have any built in loop to collect those inputs.
I'm focusing on Chinese (Mandarin) right now, because that's what I've been learning, and the language learning community on reddit likes it too. But other languages are also available.
Link: https://lingolingo.app
Right now it works on both CPU and GPU (both AMD and NVIDIA) and is capable of running LLMs like Qwen, I'm currently implementing a native profiler to trace CPU and GPU kernels and then I'll work on speed. Goal is to be competitive with PyTorch eager by the end of the year.
Source code: https://github.com/nirw4nna/dsc
My original HN post: https://news.ycombinator.com/item?id=44310678
It's never easy for me to compile this monster
Static-PIE binary with minimum options is a whopping 15M
It just keep growing
This is probably third (or fourth) incarnation of the app and I like writing a web apps with Hotwire more and more. Especially leaning more on <turbo-stream> features removes frontend complexity compared previous versions. Highly recommended, both Hotwire and rewriting your apps!
With Claude Code I really like being able to multi-task but right now it's a bit like a Tesla on autopilot needing your hands still on the wheel. With TalkiTo I can do housework/go out for lunch and keep it on the right track remotely.
It's an environment for open-ended learning with LLMs. Something like a personalized, generative Wikipedia. Has generated courses, documents, exams, flashcards, maps and more!
Each document links to more documents, which are all stored in a graph you grow over time (very Obsidian-esque).
App is alpha quality but working: https://github.com/davidventura/firefox-translator
Trying to figure out F-Droid publishing..
It's been tough to find work, so I figured I'd revisit an old SaaS idea. I worked for a home inspector in the past and saw a need for better (cheaper/faster/easier) report software.
Even if the business side flops, I'd still be satisfied with the experience. I've learned a lot about new tools like WASM and web components. I also like the UX challenge of designing for inspectors filling out reports on their phones.
We focus on people with relationship issues, and so far it's been deeply fulfilling. So many people have written in about how this has helped them heal.
Launched with React Native about 8 weeks ago, and continuing to grow fast. This is a niche space with lots of potential over the next few years I think.
Just submitted an update to help people compare their unique relationship needs to others which is so cool.
Using AI for auto subtitles and actor matching. Will build some auto deploy fragment to social media as well. I think these short fragments will do well op TikTok.
Git hosting for async teams that supports versioned patches and patch stacks instead of pull requests. All done using the standard git SSH protocol, so no git-send-email needed.
Overall, it is ending up being the most amusing thing I was working on.
edit: Very much for personal use. I currently have no intention of sharing it anywhere.
This month I'm improving CI/CD for e2e testing across Windows, macOS, Linux and Android. Also adding support for unlocking password-protected PDFs and Word docs and improving OCR. OCR runs in the background and leverages native OS OCR where available and a pure LSTM Rust implementation elsewhere. Generally improving the word processor and looking for speedups. Adding a cross-platform spellchecker leveraging native where possible, too.
Play with it online: https://tritium.legal/preview
Download for free: https://tritium.legal/download
Basically a productivity tool for making sense of reality and living your best life.
I love making something truly valuable and it's also a crash course on AI product/app development. Absolutely having the best time in my life and am so grateful to be on this path!
Building on Flutter with iOS and Android coming this summer. Desktop and Apple Watch soon after.
Full fleet/driver management platform for private transportation companies (busses, limos, white glove taxis, etc)
We just released our first B2C model, check it out at https://gouach.com :)
See https://github.com/chazapp/o11y.
These last few days I have decided to try getting Kubernetes Gateway API to work, using the implementation of Istio. I have written an `auth` microservices which provides JWTs and published a public JWKS endpoint, and intend to have the API gateway validate tokens and claims to allow access to other services. The plan being to write API services without any knowledge of the authentication systems that happen upstream. If a request reaches them, it's that it had been validated before !
- an AI Web Agent that autonomously completes tasks, creates datasets from web research, and integrates any APIs/MCPs – with just prompting and in your browser!
https://chromewebstore.google.com/detail/mkmcjgbighammmaoohb...
mmarian•4h ago
Might start doing a few posts on Cloudflare WAF as I've been working with it extensively lately. Maybe it'll help me uncover some startup ideas in that space.