2025年7月18日 的 Show HN

We launched marimo [1], an open-source reactive Python notebook, last year on HackerNews. Today, the most popular recent feature request in Google Colab’s issue tracker asks for marimo support in Colab [2].

Rather than try to convince Google to replace their notebook with marimo, we decided to just build our own cloud-hosted notebook service instead. We're calling this molab (mo for marimo), and we're launching it today.

You can try it at https://molab.marimo.io

Some features:

- Persistent storage

- Link sharing (notebooks are public but undiscoverable, like secret GitHub Gists)

- Download notebooks to your machine, reuse them as Python scripts or apps

- Upload local notebooks to the cloud from our CLI (coming soon)

- Real-time collaboration (coming soon)

- Configure computational resources to obtain more CPU or GPU (coming soon)

marimo is a modern notebook for modern data workflows; we also built molab on a modern tech stack:

- Notebook dependencies are managed by uv, enabling lighting-fast package installation (thanks to uv’s cache and more generally its performant implementation). uv makes it easy to run molab notebooks locally, too: uvx marimo edit <notebook-url> brings the notebook down to your machine.

- Persistent storage is powered by R2, Cloudflare’s zero-egress object store.

- We use Pydantic logfire to monitor our deployment.

- While our implementation is agnostic to the compute backend (stay tuned!), we’re currently running on Modal for fast startups (not to mention a slick developer experience). Modal sandboxes make it easy for us define containers at runtime, containing notebook code and their dependencies. (Shout out to Eric Zhang from Modal for helping us get started.)

molab is free to use, as long as usage is reasonable. Our goal is to make is as easy as possible for our community to use marimo notebooks.

Finally, learn more at our announcement blog: https://marimo.io/blog/announcing-molab

If this interests you, please give molab a shot and please share feedback — here or on Discord (https://marimo.io/discord).

P.S. This is not our commercial product, this is really just for our community.

[1] https://github.com/marimo-team/marimo [2] https://github.com/googlecolab/colabtools/issues?q=is%3Aissu...

I've been working on librari.io for the past several months and just launched the beta version.

The Problem: I have 500+ books across multiple rooms in my house and was desperately looking for an app to manage them properly. Most library management apps are either too basic or designed for institutional libraries with rigid workflows that don't fit personal use.

What I Built:

- Multiple libraries: manage collections in different locations

- Location tracking - remember exactly which shelf each book is on

- Loan management - track books you've lent to friends

- Custom fields & tags - store any additional book info the way YOU think about them

- Reading progress tracking - dates, duration, personal ratings

- Modern UI/UX - clean & actually enjoyable to use

Current Status:

- Beta version live

- Working on improving the responsiveness of the app and addressing initial user feedback

Would love feedback! Especially curious about:

- What features would make YOU actually use a library management app?

- UI/UX feedback always welcome

- Any book collectors here who'd be interested in beta testing?

Looking forward to your thoughts! Thank you in advance.

MockFedCM is a free FedCM Relying Party (RP) and Identity Provider (IdP) for testing FedCM integrations. Simulate real-world authentication flows, debug your implementation, and validate user experiences—all without needing a production IdP or RP.

Couldn't find a reliable, free place to share & rate AI prompts so I thought I'd take a stab at it

Already has 500+ prompts generated by AI using the latest model prompting guidelines

5 different supported prompt types: full prompt, enhancement, template, system, chain

20+ categories: coding, writing, marketing, business, creative, etc.

Every prompt gets evaluated automatically by multiple AI models (Claude 3 + GPT-4 Mini, more to come)

Then humans can rate and there is an overall score that takes both AI & humans into account

AI eval prompt here: https://www.josh.ing/ratemyprompt/evaluation-prompt

I'm still quite the noob when it comes to AI stuff so I'd love feedback about RateMyPrompt and ways that it could be improved

Driven by Mediabunny and WebCodecs.

I wrote a tutorial on how to create Bash scripts, where the command line interface runs entirely in the browser using v86 (https://github.com/copy/v86), and the code editor uses Monaco.

Author of Mark Notation here.

Mark is a unified notation for both object and markup data, combining the best of JSON, HTML, and XML with a clean syntax and succinct data model.

I'm glad to announce the 1.0 Release of Mark. This release is just the start of a long journey to make web a better platform to store and exchange data.

Your feedback welcome!

Hi HN,

UML diagrams were meant to standardise system design, but they’ve become outdated — slow to build, hard to change, and rarely used in practice. Most tools feel clunky, and diagrams often go stale fast.

I’m building RapidChart, a tool to bring diagramming back to life — fast, flexible, and developer-first.

What makes it different:

Cursor-style interface for rapid diagram creation

Fully customisable and exportable

AI-assisted (works with GPT-4, Claude, etc.)

100% free

This began as part of my research at Imperial College London, and I’d really value your feedback.

Here’s the link: https://www.rapidcharts.ai

Thanks!

Hi HN

I built *fluent-state*, a lightweight React hook that replaces useState, useReducer, useMemo, and useEffect.

It gives you:

- A fluent, proxy-based API: `state.user.name('Joe')` - Immutable updates without spread logic - Automatic dependency tracking - Built-in `compute` for reactive, memoized derived state (no useMemo needed) - Built-in `effect` for reactive side-effects with automatic dependency tracking (no useEffect needed) - Zero boilerplate, full control

Designed for local state, nested structures, and clean, declarative code.

Demo: https://stackblitz.com/edit/vitejs-vite-uh4kyzkf?file=src%2F... GitHub: https://github.com/marsbos/fluent-state NPM: https://www.npmjs.com/package/fluent-state

Try it out plz.

For folks working on Graph RAG and trying to get LLMs to generate Cypher queries, I ran some experiments on the LDBC dataset and wrote a blog post about it (code is available in the link shown at the end of the post). I've been trying to answer a burning question of mine that I've had for a while now: when doing Text2Cypher, are LLMs better at interpreting graph schemas in JSON, XML or YAML? (Spoiler alert, the format barely matters, it's all to do with context engineering and retaining only the relevant parts of the graph schema in the prompt). Results on the latest LLMs are really good!

The post also contains some other tips on graph schema design: I think we're in an age now where we need to design graph schema for both LLMs and humans. If you're working on Text2Cypher in any way, hope some of these ideas and experiments are useful!

Local LLMs now supported. Works on Windows, macOS, Linux.

Another news on Roceet. Last time I posted I mentioned Roceet (now Roceet Share), a file sharing platform as a solution to transferring data between devices. It was motivated by seeing my fiancee struggling to transfer data between her phone/laptop.

I found another pain point: she’s a wannabe model, and the process of feedback and selecting photos is a mess. Big folders, endless emails, no good feedback tools.

So I built Roceet Review. Upload photos/videos/whatever, share a review link, clients can approve pics, comment, even draw on the photos if they want to be specific.

Its kind of a pivot from my original idea about Roceet. But Roceet Review is an extra tool on top of the offer. No extra cost, same architecture, so it didn’t take forever to build.

Any feedback is welcome! Use YC100 for 1 month free. Whoever gives me feedback, I'll give you the product for free forever. I really need help shaping this.

link: https://roceet.com Old post: https://news.ycombinator.com/item?id=43754748

I had an idea for a game that fused my favorite time-waster games, Wordle and Sudoku, so I had a go at making my own daily number puzzle: Numbl

Basically, you need to fill the rows and columns with a unique set of digits 1-9 where each row or column has a constraint (sum, all odd, all even, range, etc.). It generates a puzzle each day, but you can try a random puzzle by pressing "New Game".

At work I'm easing more into using AI in my workflow so I used this as a testbed for incorporating it more (i.e. rapid prototyping and validation) without outsourcing core design or compromising engineering thinking.

I ended up with a finished product that I actually want to keep maintaining so I thought I'd share and invite any constructive advice or criticism the community would have!

What it is Symvol turns any tech doc into AI-narrated video in ~1 min. Every code block stays live-viewers can click-to-copy and run snippets while they watch.

Why we built it We began as a generic “text → video lesson” tool. Early testers asked for coding support, so we added a code-aware pipeline. This is the first public launch of that version.

Try it Drag-drop a PDF at symvol.io (free to try) OR install our Chrome extension to run Symvol on any web page

Looking for feedback on 1) Narration accuracy 2) Clarity of visuals 3) Usefulness of the live copy-code feature

We’ll be around all day - happy to answer questions or dive deeper in the comments!