alt Hacker NewsAsk HN: What are tools you have made for yourself since the advent of AI?
Comments
I have a telegram bot, and I used to have it read message in a group chat to detect the language, and if it wasn't English, it would use gpt5 nano to translate.
To save on inference, I had Claude create a lightweight typescript library to handle translations and detection. Now I get the previous functionality for 200mb~ ram versus paid API credits.
I also built a home assistant menu bar in osx, let's me easily monitor activity, toggle devices, view cameras, etc.
1. Built an agent memory tool since all agents and clis are dumb and don't remember anything. Instead of prepping 300 project files and Md files I just say:
Check sugar memory for the latest thing we were working on.
2. The second thing is when making changes across a large codebase agents are also dumb at figuring this out and also grep 300 things, using tons of tokens. Instead I say
Check RemembrallMCP to analyze the impact of the change.
TONS of tools. Most written in Go. Several have API servers in addition to cli or TUI or web interfaces. The API interface to my apps makes LLM-driven development much faster.
https://github.com/michaelteter/docgen : create a single text file of your entire project, with a tree and some other useful bits. This is good for dropping into an LLM or research notebook instead of giving an LLM access to your actual project folder. It also can be put in your pre-commit script so you always have one single doc you can diff from one commit to the next.
md2pdf: markdown to PDF, relying on defaults and optional config files or cli args for formatting choices (such as page margins)
md2gslides: markdown, converted into slides, and using Google Slides API to generate the doc in my Google Drive. This saves me so much effort (I teach, so I make lesson plans/presentations all the time).
get-music: TUI app that lets me search Youtube and easily queue up to download one or more of the search results. Then I take the downloaded content, split out the audio, LLM process the video title, add metadata for music, and then provide an easy command interface for local searches and playback of downloaded content.
bookmarks: TUI for slurping all the URLs from my browser, LLM-tagging each url based on the tag list I provide with the prompt and url, and lots of features for managing priority, show/hide tags, etc. This was to help me stop worrying about having a hundred tabs open. Now I can just sweep them up into my own private, encrypted (sqlite) db.
ESL-Planner: Complete web app for building class plans for teaching English (based on params, such as student age range, skill level, specific teaching language (what we want to teach), etc. It's close to being ready to productize and release as SaaS, but I built it for myself initially.
Numerous other tools plus a guide doc listing all the tools and what they do. These resources are then made available to LLMs when I'm developing, saving me (and the LLM) the time of hand-crafting the same tooling over and over.
A Few: - Augsentric [https://www.augsentric.com] - probably my biggest time/AI sink - for evaluating websites - FencePost - [https://github.com/seriocomic/FencePost] - a UI for multi-host Firewall rules (UFW) - EventFeed (private repo) - a timeline of events on my network in a centralized UI - Ledger (private repo) - personalized finance ledger using bank statements
The biggest unblock remains the tools/scripts/skills for documentation (started with Notion, network now sits on Obsidian for read/write).
- Email triage
- Meetup alert for meetups that match specific topics
- A daily journal that transforms entries into chibi-style cartoons
- A cashflow forecast our stupid accounting software can't do on its own
- DIY service monitoring for a ragged collection of docker containers, cron tasks, scripts and various others
Some things I’ve used AI for the last year or so:
- small club website: https://www.kolibrinkpg.com
- ticketing system with Stripe payments and QR scanning at the door
- Instagram/media ingestion for the club site
- genealogy tool with GEDCOM import
- scripts for downloading/archiving public-domain film material
- playlist/library tooling for DJ use
- music collaboration/sync tool for Ableton projects
- normal work stuff in a much larger existing codebase
I have become a lot more strict about process after being burned a few times. Mostly: make the change small, be clear about what it is supposed to do, check the assumptions before coding, use tests/logging/manual checks as evidence, and don’t merge anything I can’t review and explain myself.
Still a work in progress but I’m making my own subway web app. A by-product of this is this realtime subway map of NYC
I've made text editor engine for .NET 10 built in C# with no UI dependencies. Includes a piece-table buffer, multi cursor editing, syntax highlighting, code folding, snippets, TextMate grammars, diff engine, undo/redo, C# REPL, plugin system. Its essentially a deterministic execution runtime for structured, auditable text mutations thats built for AI systems, automation pipelines, and tooling that needs to manipulate text documents. https://github.com/marinusmaurice/TextApi
- a sky shader with the "correct" color blue, sunsets that please me, and an astonometrically correct year round sun path
- github clone + extras
- a stack (FILO) based task manager / TODO list
- a CAD kernel with Blender frontend (WIP)
- a minecraft mod that makes real terminal emulators in block form
- ^ that but in Godot + a terrible "game" world (WIP)
- a somewhat failed app organizing claude workspaces
- a somewhat failed attempt at a VM framework for MacOS
- a somewhat abandoned gmail clone
- a farmland pricing model + maps etc.
- partially reverse engineered VCDS device
- a likely novel fractal system I need to work on some math to publish
- NTSC transmitter/receiver in gnuradio for the artful corruption of video
- backend for iOS appstore handling of account/subscription things
- an RSS / Podcast reader
Recently I made Vocast (https://github.com/cnrmurphy/vocast) - a cli driven tool that uses local TTS models to convert articles to "podcasts" and expose them via RSS feed. I wanted a way to listen to articles without having to pay for an app. I convert the article on my PC (which gets added to a managed library), run a web server, run Tailscale on both PC and phone, then I can use a podcast app to access my library. Nice way to consume some articles while out for a walk or anything else and has worked reasonably well for me so far.
After I tried and failed to find any decent QR code generators online, I made one: https://www.cutearr.com/
Runs entirely in the browser, no tracking, no analytics, no ads.
I tool which pulls all details from my Garmin and keeps a '5 day rolling average' series of stats -- my lifestyle means I can't keep up 'X steps a day', but I feel I can keep up 'average X steps a day for the last 5 days'. What's perfect is I can tweak it, I even added a special 'the weather is truely awful today, you can dip under average but you need to make it up later' option by getting it to automatically pull the weather forecast, or 'The weather is perfect today, really push'.
I have built several tools from general purpose agents to the current project I am working on which is AgentLine - Telephony services for AI agents it gives a phone number to every AI agent to call and sms. I built it over a period of 14 days now I have refined it so much that I have more than 36 paid customers with 300$ in Monthly recurring revenue.
I made mobile friendly agentic driven IDE (as a PWA), where I can start "agent" processes on my different laptops (company, personal) or on the dedicated server and manage them from any browser / mobile device enywhere. Now I spent most of my time do programming in my head walking somewhere and talk to agents / review code directly from mobile. I also added support of n.eko like remote browsing streaming for emulating desktop browser while I'm on mobile and simplified deployment tool to manage kubernetes / docker deployments.
Recently I got AI to help make a script to convert a CSV dump of trading activity on a trading platform for uploading into another stock trading analysis platform.
Later on, I managed to crunch the numbers in the script, and realised I could dump them out and display them in a dashboard alongside the trade activity. So I built that too,
An excel spreadsheet could have done the same job as the dashboard, but the script for conversion was greatly aided by the AI tool. The work otherwise would have been a bit of manual coding and back-and-forth testing.
https://docdiagram.com - A tool for creating technical diagrams. Basically an upgrade from mermaid.
https://anomalyarmor.ai/ - A tool for managing data observability and data anomalies.
Used daily customers but also for my own work. These tools fill a gap that helps me get sh*t done.
I had a portal/app with messages exam results, calenders for .) kindergarten for my youngest child (who will now go to elementary school) .) elementary school for my middle child .) daycare for my middle child .) high school for eldest child
All spamming with messages and news - so I made python scripts that use the apps API to get the messages (exams and changes etc) and puts it into our family telegram group chat if relevant
I just can't sit here every morning 20 minutes on my phone logging in checking everything just to find out the food plan has changed ...
I've been building a tool aimed at better web annotations for teams and AI collaboration at https://viewall.io/
Having worked with web facing teams there are always vastly different methods any individual uses to capture their feedback. If you or anyone you know on Mac that has 100s of screenshots on their desktop, this is aimed at bridging the gap.
Clipboards are optimized with context for LLM markdown ingestion and for use in work suites like Jira/Confluence.
Still fairly early, but I've been using the tool to help build the app itself which has been an enlightening experience.
Heaps, most recent is just a little applet that stops my Mac from going to sleep with the lid closed: https://transitivedev.gumroad.com/l/doppio-app
Bunch of security tools: Some are at https://diffsec.dev others:
HN Chrome Extension for dark mode and a few other styling tweaks I wanted.
Jira Chrome Extension to add some notes and links to dashboard pages that I wanted for ease of use.
Small application which takes a CSV and turns it into a Registered Server List for SSMS in order to keep my list of servers updated for queries across all our databases.
Honestly one of the good use cases for AI. Small low complexity scripts and tools for assistance is a great use case. I'm amazed at the folks that are doing huge monolith rewrites with Agents and such, but I've never had good results for that. Small time saving scripts gets me a much more direct return on investment.
A Javascript framework called places.js for creating interactive UIs using web components. It has support for cross component state management, backend data fetching, and web scraping protection. https://codeberg.org/createthirdplaces/places-js/src/branch/...
Here is a website I made with places.js for DC area board game events. https://dmvboardgames.com/
A vocabulary flash card app. I know I know, there's a million.
None of them had share extension feature so whenever you are reading on the web or in a book you highlight the word share directly to the app and it automatically looks up the definition.
Also added a voice review (with gemini native audio model) so you can chat your reviews.
https://apps.apple.com/us/app/vocabulary-flash-cards/id67750...
Skills for creating good and repeatable benchmarking scripts.
A knowledge base for my research area, with tools for paper ingestion and search.
An md file to html presentation tool, there are several but this one helps me.
A review tool that splits a PR or branch intelligently into modules, and does per module reviews and global reviews for different aspects, and then summarizing that into a report. Can be used with multiple different harnesses. Written as a Python project, but build-time assembled into a single-file Python script with uv run --script shbang line.
Two that I use pretty frequently:
1. https://shopmath.app - I got tired of converting decimal to fraction for woodworking/renovations, and I wanted to round calculations to the nearest 16th, so I prompted this.
2. https://youtubetimestamps.app/ - I use davinci resolve and wanted an easy way to convert my EDL markers into timestamps for youtube.
Claudhd
It's a user daemon that runs on my machine and exposes a unix socket, and then a bunch of hooks in claude, zsh, vim, etc, that report directory and commands I've run and all that, pipes it to claude Haiku for summary, and then stores context in sqlite. It also exposes that data as MCP so I can use claude to say "hey what was I doing yesterday," or any arbitrary time range.
I find that in the age of using AI agents, "Wtf was I working on yesterday" is an even harder thing to remember for me, so this helps me kind of track everything with a database that a) has AI summaries already and b) can be accessed by AI as well as a CLI.
I've been using AI to build and set up two websites so far one in the travel space and the other in the real estate space. I have literally not touched code or written one line of code and brought up this to websites. Triporacle.app and landradar.land
I've been making a set of local- and files-first iOS apps for myself - Photo Gallery, Music and Contacts.
I use Linux on my computer and an iPhone and wasn't happy having to use a cloud to sync my data between my devices. Syncthing/Synctrain + my apps allow me to keep everything on my private network, without ever touching a server or having to self-host something like Immich.
I was always experimenting with different agents and models, and I wanted to track how much I use eg codex vs opencode vs hermes, and which models I use.
So I made, kind of last.fm/waka-code for agents where I track (anonymous) usage per project
here is the example of my profile: https://clankerlog.ai/kodisha
Data is fully anonymous, all I collect is (agent, model, project name [can be mapped to something else to hide the true name]) and everything is opt-in by default (per project)
I made a home dashboard (that I access on my iPad) that talks to my Philips Hue bridge for controlling my house's lights, and also shows train arrival times + what my dog should be clothed in (if at all) for her walks (there was a smallish period of time here in nyc where one day you'd have to wear a jacket and the next day you'd be sweating even in shorts). It also shows our family calendar (wife + self) - all family birthdays in the next 4 days, any events, etc
I have a service that runs on my home server and uses the Ubiquiti API to detect when I'm on a zoom call. It then pauses my Sonos network-wide. Unpauses when the call is over. The next step of it is to have it's own podcast library, so I can have my own little talk radio going on in the background all day.
Some point, I want to convert into a little app and open source it, so you can install on your laptop so it's usable by more people. And have it detect more than just zoom.
Made "stop localhost" publishing for vibe coders called Revdoku. It is useful when need to quickly publish a prototype, report, set of documents and can be done right from Claude or Codex.
If you use ChatGPT then can say it is more like ChatGPT Sites but also works in Claude.
So many! I manage a fund that buys small e-comm brands, and at this point the whole thing runs on a combination of AI and tools created with AI. My favorite is one that scrapes my Alibaba/WeChat/WhatsApp/email supplier convos daily and uses that to build a dashboard tracking the status of my orders.
I write a Substack about the whole thing and have a pretty comprehensive list here: https://theautomatedoperator.substack.com/p/15-ways-im-using...
A web harness for another open source project (CHIRP) which lets you program channels into all kinds of handheld radios (HAM).
A JS image pixelator: https://kremerman.me/pixelate/
Can be used to resize images, but the main purpose was pixelation for a game I was making.
Been working on this spaced repetition app / incremental reading app. I used SuperMemo for about a year straight, found the UI not that great, so I decided to make my own incremental reading app. Think: incremental everything. Supports multiple spaced repetition algorithms FSRS-6, SuperMemo 18, SuperMemo 20. https://github.com/melpomenex/incrementum-tauri https://readsync.org
I host a couple of services on a box in my network. I built a tunnel that runs on the box and on my VPS to allow me to quickly access those services.
A tool that checks for new movie and tv releases, looks up ratings to see if they are worth adding to my plex server (see above about services I'm running), and then finds the magnet link and downloads them. But will only do so if my VPN is connected.
A tool that allows me to quickly build out paintball fields using my STLs of bunkers that I made, and export the full field layout as a single STL for quick painting, slicing, and printing.
A github client / dashboard that can pull 20 of so repos for all internal and client projects in one UI so I can stay on top of project delivery and long standing bugs. It has global search, bookmarking and text based / minimalistic ui for maxium space utilisation and information density. It's read only so to comment on issue i click a link to open GH in new tab but helped me a lot to have this birds eye view on my company. Don't get me started on GH Project. I tried Linear many times but multi project / multi repo is just not their core focus and it shows.
A flight search tool that takes a set of origin cities and finds the cheapest shared destination: https://flightjive.com
I've created lots of small things, but the most interesting is a firefox plugin that detects when some media is playing and send an event with the details to my home assistant. This way I can create some automations that automatically change the lights when I play/pause some media. It's really cool to have your lights automatically dim when you start watching a movie and then for some accent light to turn on when I pause the movie to go to the bathroom.
transitioned from my OmniOutliner 3(!) based system last year finally to modern macOS and obsidian.
1 year later, with no js/ts skills at all, i got 10 custom plugins, several forks where i fixed bugs and some custom adapations, dozens of scripts and snippets and what not
Now obsidian works for me like I want, and still every day I use it Im still in awe
It is a bit hard to describe until I make a video but basically I seem to have changed the fundamental unit from a document to a chunk and it is just awesome
none of which would have been possible without AI
I wrote NerdCalci (https://github.com/vishaltelangre/NerdCalci), a free calculator app for Android.
Besides, I made a lot of automation scripts (mostly using Ruby) that run on my raspberry pi to fetch/parse/crunch things and notify me on my Android phone through a self-hosted https://ntfy.sh server.
- Day logger quarterly goal management and daily goal tracking system with multiple checkins, voice transcript task dump, jibberish to apple reminders, daily recommendations based on activity and goal tracking and always on dashboard.
- Snubnosed mandarin app. Vibecoded anki and tinder-like character game for mandarin which allows new vocab to be added on the fly. Also accurate text to speech for tones.
- What did I learn? Tweet summarize that takes all favored tweets and assembles into weekly categories and allows deep research on certain topics.
I am building my self-hosting llm-wiki system (https://gist.github.com/karpathy/442a6bf555914893e9891c11519...). My approach is to start with a theory of how such systems could work. Then since llms can interpret theory - this theory becomes an executable llm-wiki system itself.
It's called Commonplace: https://zby.github.io/commonplace/
I made a Safari extension with Swift that automatically suggests using Fastmail masked email addresses on login forms. Never published it, instead just using an xcode dev build on my phone. Works flawlessly.
I built a language tutor iOS app for having voice conversations with a patient AI tutor. There’s plenty of services doing the same thing but I didn’t feel comfortable sending my broken Danish speech to a startup. It’s open source, going through App Store review now. https://github.com/challenga-org/openlanguage
I built a color palette tweaker very specific to my OCD needs:
https://archives.fifthrevision.com/color-generator/index.htm...
I also have a local zsh autocomplete macro that let's me type things like "git rename annotated branch" and ctrl-g and it will get me the actual command. There's also a ctrl-r mode that searches my history using natural language. This is connected to a locally run ollama so my keys don't leak.
I’ve built a Postgres monitoring tool: https://dlt.github.io/blog/posts/pg-glimpse-postgres-monitor...
And a MCP-powered error tracking rails engine: https://dlt.github.io/blog/posts/mcp-powered-error-tracking-...
https://github.com/edumucelli/docking
A Linux dockbar with tons of applets and support for x11 and Wayland. Works on Gnome and KDE. I always wanted to write one as I have been involved with several open source ones, but it is a lot of time to go from scratch. I use it everyday, and I am enjoying it so much!
I made a little script that takes all images in a folder and formats them for my digital photo frame. It knows which have a converted file in the dest and skips them.
A more powerful search tool for the contents of Palo Alto firewall release notes: https://firewallissues.axvig.com/
A tool to periodically sync Device 42 data to Netbox (work).