Ask HN: What are tools you have made for yourself since the advent of AI?

I need to finish off that blog post.

With Tailscale, you can basically point a domain to the FQDN of a machine you’re sharing with people and the domain will simply work for them (and only them)[1]. But for it to work without them having to know or specify the specific application port, you have to grant them access to 443 (and 80) in your Tailscale ACL for that specific host.

So yeah, now immich.familydomain.com works without family members having to remember the specific port. BUT, serveradmin.familydomain.com (another app on the same host) will ALSO be accessible to them (from a networking POV). We opened port 443/80 for that host after all.

I took a few hours with Claude back in January (?) and we wrote a tiny Go authorization gateway which basically consults both Tailscale’s public API and Tailscale’s `localapi` and returns the appropriate response to Caddy based on the requesting user’s actual allowed ports.

So now I can share different apps (subdomains) with different people without forfeiting access controls, all driven by Tailscale’s policy file.

(I hope I didn’t mess up the crux of it, pretty late here)

Edit: why not (something like) Authentik? Quoting from the draft:

  I contemplated this for a bit but one thing I kept thinking about is:

  _They're already logged in to Tailscale_

  Why do I need to install and configure another full-blown app dedicated to doing something Tailscale already did? Why have the users go through two hops of authentication?”

I've made a number of ceramic molds for slumping fused glass into bowls. As well as wooden templates for ceramic mugs. I've devised a few carrying tools to move glass frit paintings from my studio down to my barn where the kilns sit without spilling the glass.

Or were you only asking about digital tools? I haven't really made any of those. Making physical tools feels much more satisfying these days.

Created a few Open Source tools:

Open Camera Control https://github.com/jcubic/open-camera-control - that allows me to control the settings of my DSLR while I'm recording myself.

Horavox - https://github.com/jcubic/horavox - A speaking clock

Mutimon - https://github.com/jcubic/mutimon - a config driven web scraper (found this post from the email sent by this tool).

ASCII-Globe - https://github.com/jcubic/ascii-globe - JavaScript library that renders a rotating earth or any map. Can be used to add animation to your website.

I built a browser extension that stops animated images such as GIFs by default.

I've always found it annoying that browsers autoplay animated images, and there still isn't a built-in way to control that behavior.

The extension shows the first frame and adds a play/stop button directly on the image.

What started as a personal utility ended up being published on the Firefox and Chrome extension stores. It's still a small project, but it solved a problem I had every day.

Firefox: https://addons.mozilla.org/en-US/firefox/addon/gif-control/

Chrome: https://chromewebstore.google.com/detail/gif-control/nhoihin...

Bug reports and feedback are very welcome.

Disclaimer: I previously posted a Show HN about it, but it didn't get traction [1].

[1] https://news.ycombinator.com/item?id=48208916

It's dumb, but....

I made a scraper that searches through all the news and finds anything 'war' related. It then summarizes it and give me that new in the Star Wars crawl with music.

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).

A pile of various tools:

A self hosted web archiving tool with support for extendible processing pipelines (eg. extract article -> translate -> summarize -> generate tags, download video -> split audio track -> transcribe -> summarize), which led me to make a managed chromium browser with extensions and warc support for archiving, and a RSS feed synthesizer (take random article listing page that doesn't have RSS and generate a feed for it) so that I can plug it into my archiver. An active learning loop for a model to clean up articles by removing junk like native ads and sponsored blocks.

A tabbed terminal with project management features like launching the database, app server, and claude code in different tabs with one click, and split browser/terminal panes (eg. opening a browser automatically at the correct URL when the terminal reads http://localhost:4000/).

A modular MCP server with a MCP proxy and OAuth2 dcr so that I can easily add new random ideas for MCP servers in a few minutes with Claude and deploy them such that it's available to Claude by refreshing the tool list.

A small tool to render Claude conversations so that I can link to them from my obsidian vault with something like convo://claude-code/-home-jfim-projects-foo/<guide>

And overall just deploying docker containers for my self hosted setup

Most of it is on GitHub, in various states of readiness.

I wrote my own Claude Code agent. It leverages some parts of multiple different agent stacks but using my architecture. It's layered, has long-term memory, agency to expand upon work, access to linear, access to all the models and endpoints you could possibly wish, as well as support for combining multiple providers and models into the layers so that delegation to the workers happen on the cheaper, more cost effective models. All of this runs at scale in an orchestration platform I wrote, using said agent, to create a cluster of docker containers in a JBYOS configuration, cross-cloud, k8s or swarm or whatever can run docker. It's pretty sweet and is basically an AI software development shop. I only have to give it ideas and goals.

I made a golang socks5 proxy that routes traffic to different VPSes (or the default gateway) based on hostname, over mutual-TLS tunnels, authenticated using ed25519 keypairs shared out-of-band. The "client" and the "server" are the same piece of software, and there's a web UI for configuring the routes.

I made it to deal with internet censorship in the UK, where different sites have different optimal exit jurisdictions, and most sites work fine so I can avoid the extra hops where possible.

It also works well for video streaming sites with geofencing, since the geofence itself is usually implemented in the frontend rather than the CDN. So only the frontend traffic needs to be proxied, while the bulk CDN traffic doesn't need proxying at all.

Socks5 is the ideal layer for hostname-based routing, since the proxy can see DNS names without needing to sniff TLS SNI (which is incompatible with ESNI/ECH etc.)

iirc it was basically all done in a single prompt, and I've been using it ever since. The only issue I've encountered is with WebRTC not working properly with some services. (Presumably it breaks the NAT holepunching process or something, I never diagnosed it)

Another project that isn't quite finished is a "universal" web video downloader that works by shimming the MSE APIs and remuxing the streamed segments back into a regular video file. The idea is that if you can watch it, you can save it - including but not limited to youtube videos.

I finally shipped a Chinese learning app I wanted for myself for ages at https://wenmoji.com/. Just never had enough time to sit down and code it end to end. Still need some improvements of course, but will slowly chip away on it. I use it daily/weekly myself now.

Highlights are that it completely free, no login required and works offline (once you "downloaded"/cached some files the first time around).

I’m working on Descartes[^0]. First to help diagnose what’s wrong with a machine. Later to help manage and monitor it by letting an agent build layers of tailor-made deterministic rules and statistical models, a bit like the description of the ship’s AI at the beginning of Absolution Gap. And serve as an ops "point of contact" for other agents for the machine / fleet of machines it’s in charge of.

Sometimes I daydream that the end goal demo would be to set one up in a VM and let a sort of ransomware lose in it, and see if the agent can identify what’s going on and react in time.

But for now I’m fighting Apple notarization to enable local notifications on macOS.

[0]: https://github.com/lightless-labs/descartes

I wrote a pen-plotter GUI and gcode sender in Rust. By hand. Like an animal.

I am the only user. Sometimes it's the process that matters, and exercising your brain is important too. I get that there is a lot of existential dread around AI taking our jerbs, and excluding humans from the process of creative work, but... you can still just write code, just for the personal satisfaction.

https://github.com/armyofevilrobots/bap-egui

A voice memo app, quite like the actual voice memo app from Apple. The thing is: now I can put my voice memo's on iCloud put Claude Code on it and make my transcripts into structured notes that my app then also displays.

So basically a way to just go on an hour long walk with myself, spit everything from the top of my dome stream of consciousness style, and then have Claude structure whatever I said.

It's nice to have something that structures my thoughts by just thinking out loud.

I vibecoded it (it's approaching 20K lines including tests). It works quite well but there are some bugs, so will have to do some actual engineering. But the UX is working quite well.

Tools? Mmmm i tried making a openCV based tool to recognise circular objects on sat. imagery, it didnt work at all (tried for two weeks), changed to a LLM, it seemed to work, or it was just a little bit off, O gave up. If anyone wants to collab in this space, shout out! Ah also a translation tool for pdfs… its also stuck in limbo, forgot why again…

The most comprehensive tool I've built so far was a JSONPath playground [0] when I was working on a game where I found myself writing a fair bit of JSONPath (well, JSONPath Plus specifically) expressions by hand and wanted to be able to test them out in the same way that I would on regex101 when writing regular expressions.

I realistically probably would have only saved myself less than an hour of crafting those expressions if the tool had already existed (with this level of detail, there are lots of many simpler ones already for it), but I would have spent a solid 40-50 hours of bouncing between manually crafting and writing detailed instructions to direct the agents to get this tool there.

[0] https://jsonpath101.com/

I've made 3 that use pi as the primary interaction, with custom tools and scripts.

1. family tree based on wikitree format. Transcribe records, verify/edit, then incorporate them into the tree with full citations and biographies. This one is the big one. It includes a tree browser and best genealogical practices.

2. Pool Math replacement. Log pool chemistry tests to markdown files and suggest the right amount of chemicals to balance.

3. Calorie counter. Log calories to a markdown file, look up foods and amounts in online databases, sync with garmin connect for exercise calories.

All of these are written with AI but also are interacting with pi and telegram, mostly using deepseek v4 flash.

Built a tool to help design cs/science inspired jewelry in CAD. I wrote a DSL to describe the jewelry, and had an LLM write the interpreter to generate a CAD file using cadquery (note to self: LLM suck at 3d reasoning). I would not have had the time to do this without AI.

Also used AI to design an online store (I'm not a front end dev). It's amazing to see my wife (non-technical background) tweak the web-site using claude code.

End result: an online store where we sell jewelry pieces that actually are lambda-diagrams (Tromp diagrams) that compute Graham's number, or of the Y-combinator (well, technically it's just a fixed point operator, one beta reduction away from the Y-combinator. But Y-combinator was not aesthetically pleasing from a jewelry point of view)

store: Built a tool to help design cs/science inspired jewelry in CAD. I wrote a DSL to describe the jewelry, and had an LLM write the interpreter to generate a CAD file using cadquery (note to self: LLM suck at 3d reasoning). I would not have had the time to do this without AI.

Also used AI to design an online store (I'm not a front end dev). It's amazing to see my wife (non-technical background) tweak the web-site using claude code.

End result: an online store where we sell jewelry pieces that actually are lambda-diagrams (Tromp diagrams) that compute Graham's number, or of the Y-combinator (well, technically it's just a fixed point operator, one beta reduction away from the Y-combinator. But Y-combinator was not aesthetically pleasing from a jewelry point of view)

store: https://studio-galois.com

Many, really, but there are few I'm especially proud of:

- https://github.com/exlee/pikchr_pl - DiagramIDE (diagram amiga-style workbench where you can script Pikchr diagrams using TCL, Prolog, Pikchr or - recently mruby). Note: you need to navigate to actual crate for description. There are binaries built in case someone wants to try it.

- https://svg.axk.sh - semi manual SVG fitter so that I can easier vectorize AI-generated pseudovector images (who doesn't like 30kb SVG versus 1.8mb PNG?!)

- https://github.com/exlee/rik - this one makes me laugh every time I use it - it's an AI harness with text editor as an UI (i.e. it reacts to comment strings) - I gave it personality so it makes wacky comments but other than that it's very constrained agent (limited edition ;))

These are not vibe coded but AI made it much easier to slide through major friction points (e.g. for SVG fitter I really didn't want to reinvent fitting algorithms)

- https://github.com/sethdeckard/atria : TUI for managing multiple AI coding agents that doesn't force a particular workflow on you. You can use tmux, the built-in PTY, or terminal integrations with iTerm2, kitty, or WezTerm.

- https://github.com/sethdeckard/loadout : TUI and CLI for managing a personal library of Claude Code and Codex skills across your machines.

- https://github.com/sethdeckard/atlas : TUI and CLI (also works as "cd launcher") that creates a smart, automatic map of every Git repository under your projects root.

I'm using these almost daily.

A "Home Agent" setup, with customized special agents to manage various aspects of the house through home assistant, learning feedback from household users to try and tune everything at the right time.

Various MCPs for above.

A "remote claude code server", that gives project level overview and lets me run projects / develop on my home server rather than locally through my laptop. Supports ssh as well as a web UI (projects in a list, shell rendered using https://xtermjs.org/, with a tile overview when working on multiple projects to watch for turn ends.

Similar to above, I have a local version that auto launches a new project scoped podman container, passess through the work directory, installs CC/Codex/Grok into it and passess through the auth / config for each agent. Then dumps you into that shell with aliases to map each agent to that agent with a few special env flags to disable permission prompts, so claude = `claude --dangerously-bypass-permissions`.

An extensive MCP for Obsidian that gives agents access to use a lot of the more advanced Obsidian functionality, such as suggesting and installing plugins / configs / etc.

It's helped some of my daily productivity, but I still prefer to get my hands dirty with code most of the time rather than full prompt it.

I didn't build tools in the classic sense of something you build and run semi permanently, like baked into your setup or home server or whatnot; but I found myself building bespoke tools with most new projects at work. Get a new Jira ticket, figure out which components will be involved, often times the tool goes collecting logs and parsing them into a Web UI with buttons to toggle various features or params. And it tends to be different for every project. It's like the oldschool shell oneliners but more powerful and easier to write.

- My own whisperflow bike shed

- Converted invoicing to Typst (from LaTeX)

- Automation of blinds

- Automation of lights

- Python library to control lights

- ML tuning library

- ML feature interaction library

- Jupyter notebook slideshow interface

- Davinci Resolve Authomation

- Arduino eink bluetooth HR monitor

- Tons of small scripts

I wrote a tui sql client to replace DataGrip (which is really slow). https://github.com/keithn/sql It's quite customized to what I wanted, I haven't really checked it works with other things.... only thing is, I don't really use it much anymore, I just get claude to do all my querying.

Most of the tools I write now are bridges to various SaaS products that have APIs but no CLIs.

The tool I'm most proud of is "Hex Flex" (https://seidleroni.github.io/Hex-Flex-Web/). It is a tool to view and compare the contents of Intel Hex files. Should be useful to other people who work in the firmware field.

Not exactly a tool, but I also made pelohard.com which ranks the most recent Peloton classes by difficulty. Updated twice daily.

I've always wanted my own VW diagnostic tool suite, and between tooling that was released in public on GitHub (https://github.com/kartoffelpflanze/ODIS-project-explorer) and my own research from years ago, it always seemed straightforward but too tedious to execute on. Claude did a great job making something useful, https://github.com/bri3d/mcd-diag-rs , and now I don't have to find a Windows machine or remember a specific diagnostic cable to replace my brake pads.

I also build a ton of household glue stuff; I was never really passionate enough about the whole "homeserver" thing to spend the effort in going beyond basic video recording for my security system, but now I have all of my local-only home automation stuff wired together, mostly into HomeKit, and have been able to ditch a ton of cloud services.

I vibe coded myself a simple little Home Screen-installable webapp that tells me when my first meeting is the next day (I am in a lot of meetings, and they're constantly in flux). That way, I don't need to be logged into anything work-related on my phone, or even mentally engage in that world. I just want to know when I need to wake up. All it shows me is the time, and I can tap on it if I want to see the title. It adjusts the font and color according to how early the meeting is (earlier than 8a gets Nosifer).

Could I have done this myself? Of course. Would I have tho? Prob not.

This kind of simple, hyper-specific bespoke utility is the perfect thing for vibe coding IMO.

I'm building a replacement for TablePlus: a TableAI database client, because the latest releases of TablePlus have gone down in terms of user interface quality. You can find it on the Mac App Store(TableAI - AI database client)

I'm working on a self-hosted search service called Hister (https://hister.org/ - https://github.com/asciimoo/hister) with the goal to reduce dependence on online search engines and AI answers.

Hister is a full text indexer for websites and local files which automatically saves all the visited pages rendered by your browser. It provides a flexible web (and terminal) search interface with offline result previews & detailed query language to explore collected content or quickly fall back to traditional search engines.

It can provide a privacy-respecting search experience for serving "recall" type searches where users retrieve previously visited content, but falls short in "discovery" type searches (yet).

I made a tool that creates sandboxes (docker, podman, orbstack, seatbelt, tart, containerd, kata, firecracker) and then sets up an agent (claude, codex, gemini, aider, opencode) inside it with max permissiveness (no prompts to call sed, etc).

It creates its own copy of your workdir for the agent to play in, and then you pull changes out ala git diffs or commits.

It's a MASSIVE time saver, and I use it as my daily driver.

https://github.com/kstenerud/yoloai

scratched my own itches related to OCPP, latency, git, ssh and JMX. none involved AI

Working on a web client and bouncer for Hotline, the old chat software from ~1997. Just want to chat with my peeps who still use Hotline, from the comfort of the browser I have on whatever machine, while some server maintains the persistent connection to the Hotline server for me. Like an IRC bouncer, but for Hotline.

A MacOS desktop app and a mobile app for instrumenting GPS routes.

Screenshot here: https://x.com/LyleMakes/status/2063784301594853657/photo/1

https://github.com/Opfour/warfare - A modern HTML5 remake of Warfare 1.0 (1995) by Carric Moor Games. Turn-based hex strategy with city management, unit recruitment, tactical combat, and AI opponents — all running in the browser with zero dependencies. Playable but still building in additional features

https://github.com/Opfour/coeus-ci - Named after Coeus (SEE-us), the 100 eyed Greek Titan of the inquisitive mind — whose name literally means "querying." CI stands for Competitive Intelligence. A business intelligence OSINT tool that builds company profiles from free public data sources. Give it a domain — get back a scored report covering stability, growth, tech maturity, financial health, security posture, and transparency.

https://github.com/Opfour/op4 - Op4 is a terminal-based encrypted messaging application written in Rust. It provides end-to-end encrypted private messaging with post-quantum cryptography, routed entirely through the Tor network so that neither the content of your messages nor your IP address is exposed to anyone — not even the person you are talking to.

Spartan (Private) - Open-source safety app for women. Community-based emergency response.

AATR - (Private) Catering unified platform: events dashboard, pack lists, and staff management

https://github.com/Opfour/thelinuxreport.com - Linux news and information aggregator.

God I love this stuff!

(edit: I also have about 6 more projects I am working on locally not yet uploaded to GitHub)

RainBreak - https://rainbreak.franzai.com/

Blocks the computer for x minutes. Agents keep on working. AI doesn't need a break, but I do. And honestly, at least for me AI has made my desktop as addicting my social media feed.

I built a complete clone if fly.io infrastructure: VMs, networks, etc. so I have my Vercel on bare metal machines to maintain full-stack apps on https://playcode.io.

It took more than a year. Why? Cloud infrastructure is too expensive.

* Wikimedia Commons upload: https://daniel.lawrence.lu/blog/2026-03-25-uploading-to-wiki...

* Image viewer that can handle really big photos + run scripts via custom keybindings + CLIP search: https://daniel.lawrence.lu/blog/2025-10-22-sriv-simple-rust-...

I built a database.

The team behind Gel got acquired by Vercel and I already experienced falling in love with a dead database (RIP RethinkDB) so I decided to fork the concept to a TypeScript port with Svelte as the UI instead of React.

It'll live at disc.sh in a few months. Early dogfooding is promising.

EDIT: Also forgot that I removed React from GraphiQL in favor of Svelte too. https://news.ycombinator.com/item?id=48044888

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...

It's not just for myself, but I'm primarily creating it for myself - it's a browser for designers. I work in code but I often want a figma-type interface to explore different ideas without having to branch or litter my codebase with a bunch of demo components/files.

Normal browsers have built-in dev tools - this has built-in design tools. so I can visit my app, open up a surrounding canvas, pull fragments into the canvas, do some design-ish stuff, and merge it back into code. All in the same UI. It was cool enough that I'm going to release it, but for now it's very useful for myself.

https://matry.design/

Lookup or modify selected text using AI (chrome extension). I just select any text and click the tiny popup button "what's this" and get an answer right there on the page. Made it mainly to explain terms and abbreviations I come across on HN often. Can also ask any other question about selected text. Can even modify the selected text the same way. [1]

OneNote to markdown/obsidian canvas converter. It did that using interop api to read the actual XML of the onenote files.

Work time tracker as 1px line on edge of monitor. Shows thin line at the edge of the display which fills up based on what i am doing.

Plaintext bookmark chrome extension that save links to local markdown file, Dynalist, Workflowy, Github Gist and import export between them. Was originally for Dynalist when AI couldn't do much 2-3 years ago. Recently added these other end points. [2]

A heart rate monitor with finger on camera. It's bit crappy though. Had to make it because many trackers, including google fit, couldn't detect 200bpm. https://github.com/SMUsamaShah/heart-rate

[1]: https://github.com/SMUsamaShah/LookupChatGPT/tree/claude/fix...

[2]: https://github.com/SMUsamaShah/plainmark

- Otzel, an OT library for elixir that is in some common cases 50x faster than the most widely deployed elixir OT library:

https://github.com/ityonemo/otzel

- Spector, a bare-bones CQRS library for elixir that composes extremely well with Otzel:

https://spector.hexdocs.pm/Spector.html

- nanodrop, an elixir library that interfaces with nanodrop spectrophotometers:

https://nanodrop.hexdocs.pm/Nanodrop.html

- opengenepool, a web-based dna editing tool (technically I started this one 11 years ago):

https://opengenepool.vidalalabs.com/ https://github.com/vidala-labs/opengenepool

- a proprietary camera library that communicates with industrial-grade digital cameras.

- a full on self-hosted lab management system (journal articles, obsidian-like notes, lab notebook, kanban, embedded dna editor, LIMS), obviously this builds on all of the above items

A weight lifting app. I’ve paid for, and used, others over the years, but I always wanted to customize them in some way just for me. So, I just decided to create one the other day (used antigravity CLI) and I’m hosting it on Vercel as a PWA. I’m enjoying it so far and see a lot of potential with making hyper-personal software moving forward.

I have made a TUI tool for organizing and authoring sermons written by myself. It also has a third mode for reading one or two bible versions in parallel. I love the interface and keep adding features to make it easy to navigate in and something that actually helps me to organize and write sermons. Some features are to have notes at specific places in the bible, summaries, notes, exports to html/pdf, metadata for each sermon and autofocus on widgets when changing between the three modes. Happy to work in the terminal this way. :-)

A graphviz substitute in rust:

https://azriel.im/disposition/

The text editor has an LSP built in to guide you to what can be entered, but see the examples.

I wanted a way to have visual documentation that can show/hide detail, and be pure svg so that it can be pasted into sites without becoming pixelated.

Also supports animations to show interactions between hosts, which always gets messy when drawing a static diagram.

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 terminal-native agent multiplexer built on tmux. Similar idea as herdr but wraps tmux in an outer TUI instead of replacing it entirely: https://hmx.dev

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.

https://probplanner.com/ - I never had the time to dedicate to building a Monte Carlo simulator for project estimates. It was always something I just couldn't justify given my short commute. I used this project to teach myself how to use Claude Code and Codex over last summer.

I've done a lot of little things in Emacs since. Just minor things to improve my workflows or build up Chief of Staff type information flows.

Having fun building something in software I always pushed for "when I will find the time".

Being proud of the result.

THAT is a real game changer LLMs allowed me, both in my professional and my casual life.

For example this:

https://github.com/yodalf/coincan.git

or this:

https://github.com/yodalf/kiosk.git

I built a tiny tool to help decide the seating chart for my small wedding. It was a cute GUI on top of a simple constraint solver.

It wasn't perfect, but it helped me feel confident in the final result.