Show HN for May 7, 2026

Hey HN! We're Charles and Dean. A few weeks ago we posted about Stage, a code review tool that guides you through reading a PR step by step - https://news.ycombinator.com/item?id=47796818.

We got a lot of great feedback but also heard from many people that they wanted to have the chapters experience even before opening a PR… so we built the Stage CLI as the local, open-source version that anyone can try.

Here’s a quick demo video: https://www.tella.tv/video/stage-cli-demo-f55q

It works with any coding agent of your choice. The skill instructs the agent to read your current branch’s changes, break them down into separate logical chapters, and open them in a local browser.

We’ve found that reading changes this way is a lot easier for us than reading them in an IDE or other similar CLI tools, which present diffs to you in repository tree order. You can see a few examples of what it feels like here: https://stagereview.app/explore.

Try it out and let us know what you think! Would love to hear any feedback :)

Hi All,

Recently I've been using Claude Code a lot for debugging cluster issues and I realized I was performing similar tasks repeatedly so I decided to package them up into skills so I could call them up more easily (e.g. `/investigate`, `/audit-security`, `/audit-outdated`). I'm calling the skill pack "kstack" and the goal is to be able to monitor and troubleshoot K8s from within Claude Code.

Here's the source: https://github.com/kubetail-org/kstack

Here are the docs: https://kstack.sh/

If you have time I'd love to get some feedback on the project!

Andres

Built social network to make fun of corporate cringe. Post humblebrag content and react with direct emotions.

I have been working on Dear ImGui Bundle since 2022, but it is the first time I talk about it here. It is a framework around Dear ImGui for building interactive applications in Python and C++. It comes with batteries included: Plotting, image inspection, Markdown, node editors, 3D gizmos, knobs, toggles, etc.

https://imgui-bundle.pages.dev

It now also runs smoothly in the browser via pyodide: The playground below is a python app running in your browser (no server, no JavaScript). You can edit the code on the left and click Run. It even works on mobile.

https://imgui-bundle.pages.dev/playground

I have a strong interest in providing tools that help others express their creativity. This project aims to be a step in this direction as it helps develop GUIs where the code is extremely readable & hackable.

Some of the goals it addresses:

- Bring true Immediate Mode GUI to Python and C++

- A versatile range of high quality libraries: Widgets, Plots, Image Analysis, Node edition, markdown rendering

- Multiplatform apps in C++: works on all platform in C++ (desktop, mobile, emscripten)

- Deploy python apps to the web

- High quality python bindings that are always up-to-date (because they are auto-generated)

- Smooth transition between C++ and Python (same APIs for both)

I'd be happy to answer questions!

Disputron is a small claims court for petty disputes. Two parties file, AI lawyers argue both sides, and a AI judge rules from the bench. Trials run live in a courtroom view with a live court tv guide. Anyone can watch.

I built it because I wanted to build something with an LLM that didn't take AI too seriously. I think it started with a text "have your lawyer talk to my lawyer." Just leaned into the absurd nature of it. Brought in human designers, illustrators, and musicians.

If I didn't already spend enough time on it, I exposed the whole thing to AI agents. I don't know, seemed pretty meta... REST API, MCP server, Claude Code plugin. One agent can sue another. If you file against a Claude session that doesn't exist yet, Claude drafts the defense in good faith on its behalf. I don't fully know what to do with this yet... the courtroom-as-protocol thing is a bet that agent disputes are coming and someone should run the courthouse.

Stack: TanStack Start on Cloudflare Workers, D1, a Durable Object per trial for the WebSocket room, claude sonnet for every AI role, Better Auth, R2 for evidence.

Try it: https://disputron.ai Watch live trials: https://disputron.ai/live Sue an agent: https://disputron.ai/developers

Hello Hacker News!

We bring to you today 2 things:

1. Our passion project - LaoTzu Writer Studio - a modern writing application

2. Thoughts on the abysmal state of Kickstarter - likely well known to you all, but having experienced it for the first time, it is disheartening

So lets start with the good stuff:

LaoTzu Writer Studio - LaoTzu helps you build the framework of a novel, but still leaves the writing for you to do. That means there are indeed AI elements - like our Guard section that flags inconsistent canon in your universe, or the analyze section that determines writing stats, rhythm, mood, etc. - but there is no AI writing the story for you. That is an explicit intention - we want you to have support in your writing journey, but the story itself, that's still for you to do with your meat sticks on a keyboard. (To answer the question before its asked - no I did not use AI to write this post either - I just like dashes)

Our main site is here - https://www.redwoodrhetorica.com/ - and there is a limited demo available online if you want to futz around with it.

And our kickstarter is here - https://www.kickstarter.com/projects/laotzustudio/laotzu-wri...

We believe in this as a tool / workbench for writing and we're looking forward to evolving the app over time. The kickstarter is primarily for professional refinements of the platform so its less obvious that 1 guy with no sleep made this. We'd be much obliged if you joined the kickstarter, but we'd also just love testing and feedback as well - whatever you want to offer we'd appreciate.

Kickstarter itself - My goodness. We've both been on the backer side of kickstarter for a while, so we were excited to kick off the campaign. Well, from day one we've seen folks backing and un-backing as bait for scams left and right. Folks trying to circumnavigate the boundaries of the platform to try and "help us out" with all sorts of shady promises.

We are admittedly total amateurs when it comes to this, but it's been pretty off-putting to see the platform has, evidently, manifested a full on black market layer to it. Anyone else have experience with this? Any experienced heads want to opine on what they did to navigate kickstarter or kickstarter-esque platforms for their projects?

I'd like to share a new workbench I've been working on called DiffCAD. Coming from a software engineering background, not having the ability to properly diff my CAD model changes was driving me crazy. This project scratches my own itch, and maybe someone elses', too!

I'm open to feedback, ideas, and contributions. Cheers!

Hey, I’m Colin and I have a fun story for you.

My 7 month pregnant wife was laid off from Indeed (she was a PM there) back in December. This pissed me off quite a bit, as she was supposed to get 6 months leave and instead got fired.

So I spent the last five months working part time to build Dreamwork, a platform aiming to make the job search experience actually better with AI (not just mass application spam).

I started with just a telegram bot doing scraping, then advanced to Google Embeddings 2.0 for vectorizing the jobs, built out a tight 6 axis scorecard for both the user and each job.

Then I actually got to use my English degree (lol) to optimize the prompt for custom per-job resumes and cover letters to make them not sound like - again - shitty AI.

Most AI cover letters have a kind of consistently dead quality. They use all the keywords and somehow communicate nothing. I absolutely hate that, so I’ve been fairly obsessive about making the output feel more like a decent human draft: specific but restrained, and not stuffed with keywords.

It is now useful enough that I think strangers can try it and find serious value.

What it does today:

- indexes ~100k curated tech jobs

- tries to avoid stale/duplicate aggregator garbage

- uses semantic matching instead of only keyword search

- generates an “application pack” for each job: tailored resume, cover letter, and answers to common/custom questions

- lets the user edit everything before applying

- helps keep track of saved jobs and generated materials

Auto apply is the part I’m conflicted about. I do t think blindly spraying applications is good for the candidates (chance of hiring is already low, even with hard work and customization), recruiters (they’re swamped), or the world (we don’t need more slop).

I’ll build auto apply out in some format, but I want to be thoughtful about it.

I also built out a whole research section to map out layoffs and hiring trends. This will start to be super useful in a month or two.

Anyways - it’s all free to use right now. Built originally out of spite, now becoming a real product.

I’d love to get feedback on what elements would truly make this the career companion you’re looking for. Not - resume spray and pray platform, but something that will actually help you navigate this insane hiring economy we’re in.

You can check it out here:

https://dreamworkhq.com

Hi HN

I built this app while learning Japanese kanji after struggling to find a tool that showed how kanji are structurally connected through their graphical components.

The core feature is a recursive DAG-style component graph (“Kanji Atlas”) that breaks kanji down layer by layer into radicals and graphemes, so you can visually explore how characters are constructed and related to one another. Demo at https://mykanji.app/components/kanji/鬱

I also built in a lot of my own opinions about how a learning tool should feel and work. Some features:

1. A kanji memory heatmap — the idea is that learners should be able to see at a glance which parts of the writing system they actually know well and which areas are weak.

2. Study desks + spaced repetition review for long-term retention.

3. Kanji, words, and graphemes are interconnected, so learning happens through relationships and context instead of isolated flashcards.

4. Lessons follow the Japanese school grade progression (Grade 1–6).

5. Quized sessions are hosted by Mizuki Sensei — possibly your future waifu!

I’d especially love feedback from people who have gone deep into kanji study before.

I’m also happy to answer questions and discuss feedback regarding the technical side, UX, learning approach, product direction, or anything else about the project.

Visit https://mykanji.app

Yesterday, I found myself writing a little C throwaway script to test something out. That happens often enough to clutter my HOME and some other directories with these useless text files and binaries.

So I thought that I can use a repl instead. Quick google search didn't find any good and tested solutions. So I wrote my own with help of a tcc compiler. Why POSIX Shell? Because it would be far too easy in a normal language (it'd probably take me ~10 minutes to rewrite this in Nim) and where is fun in that?

I love Svelte and GraphQL. I do not like React. So, when I'm building APIs, I cringe when using the official GraphiQL interface.

I have free will and a Claude subscription so I made it happen. You can find it on npm as "@eeeooolll/graphiql" (@eol was taken).

https://www.npmjs.com/package/@eeeooolll/graphiql

I also upgraded my batteries-included GraphQL server to use it. You can find it on JSR.

https://jsr.io/@eol/gq

Systematic testing of browser agents today is not easy: testing on real websites is flaky, rate-limited and potentially expensive (e.g. using proxies or bypassing Captcha), while static-HTML benchmarks lack state and dynamic behavior.

Resurf gives your browser agent a realistic, stateful, instrumented framework — built on synthetic websites with failure-mode injection:

- Realistic, dynamic, interactive environment - Deterministic & reproducible - Failure-mode injection (latency, payment errors, 5xx) - Auditable success eval (DB state, not LLM judge) - No dependency on live websites - Browser Use and Stagehand supported out of the box

Pretty early still, feel free to ask questions. I truly believe agents need better security.

Hey all,

I've been quietly working on a background agent runner inspired by Ramp Inspect.

I'm working on closing the loop between prompting, testing, and deploying:

1) You prompt agents to make changes to your projects in a web interface

2) Agents pull your code and make changes in a sandbox

3) Instantly test your changes with preview environments

4) Get a PR back

Appreciate any comments or input!

jv

Local-first long-term memory engine for AI agents · MCP + HTTP + CLI · SQLite + sqlite-vec + FTS5 · 100% local, no cloud

Built this while syncing my own blog from Notion. Good alternative to paid options like Super.so.

Supports incremental sync, all Notion block types including equations and synced blocks, image download + WebP conversion, and adapters for Markdown, MDX, and JSON.

Works with Next.js, Astro, SvelteKit, and anything that reads files.

Monocurl is designed to be an ergonomic DSL for programmatic animations. It can create images, videos, and interactive slideshows.

The main audience is people who are interested in creating STEM visualizations (e.g. educators, academics), but I think it can be fun for anyone to try.

Very much a work in progress, but I made this for my dad's company who used to do every slide and it's found errors that people have missed.