Show HN for September 9, 2025

Hi HN!

I’ve always been a fan of application launchers, and I was impressed by the approach the Raycast team took — especially their extension system. About six months ago I started building something similar for Linux, aiming to integrate deeply at the OS level and give extensions a lot of power.

Vicinae is written in C++ with Qt Widgets. I chose Widgets over QML for more imperative control of the UI, especially around extension handling. So far that’s worked well — modern C++ is great.

To support my goals I built a number of custom widgets, including a fully virtualized list that can efficiently render tens of thousands of items. That gave me a lot of respect for Qt — it’s a powerful framework that mostly stayed out of my way.

A key feature is support for Raycast extensions (React + TypeScript), most of which can be installed and used directly inside the launcher (though not all features are implemented yet). There’s also a native API package (@vicinae/api) for writing Vicinae-specific extensions with additional capabilities. This required writing a custom React reconciler — surprisingly straightforward, though still unpolished.

Like Raycast, Vicinae ships with powerful built-in modules, but the goal isn’t to make a clone. I want it to grow into its own project that fits the FOSS model better, while staying compatible with the Raycast ecosystem. I also plan to bring it to other OSes eventually.

I’d love feedback on the technical approach, and suggestions for what would make this useful to you. Contributions are very welcome — I’ve already been pleasantly surprised by how quickly people started helping.

Docs: https://docs.vicinae.com Repo: https://github.com/vicinaehq/vicinae

Hi HN,

Here's a tool I wrote for retrying fallible commands. Nothing groundbreaking here, this is a tool that's been made many times (and several have been submitted to Show HN). Though this one does have a more comprehensive feature set than most. I hope one or two people will find it useful.

I wrote `attempt` for two reasons:

- To have a more featureful alternative to `wait-for-it.sh` for use in Docker Compose. Specifically to apply migration scripts to a database that may not be up yet. I wanted to be able to inspect the error messages from my migration tool & retry on connection errors.

- To test a hypothesis I had that a good way to make a CLI was to copy the API of a good library (in this case, `tenacity`). I want to write a blog post at some point to discuss this at length, but the tl;dr is that I believe it was a success.

Here are some usage examples: https://maxbondabe.github.io/attempt/usage.html

There may not be much to discuss for such a small tool, but I am open to all feedback and am happy to answer any questions.

Cheers,

Max

Hi HN, I’m Eoin, founder of Sourcetable (https://sourcetable.com).

Today, we’re launching Superagents. You can now connect your spreadsheet to any database, API or MCP server on the Internet. All of that data is available inside your spreadsheet, and you can use AI to analyze it and build models, reports and visualizations.

The reason I started the company is because I spent 10 years at startups across engineering and operations roles and realized that Excel and Sheets weren't architected for the modern information environment. This creates a tremendous amount of nuisance and busywork cobbling together SaaS tools, reporting suites, and the misery of endless coordination meetings to make it all happen. (Boo meetings!)

Spreadsheets aren’t just a business application: they’re the original thinking tool. The quality of these tools has a downstream impact on analytical thinking and creativity writ large, so this is a problem worth solving. Fast forward to today, we’re a 6 person team taking on Excel, Sheets and ChatGPT, so we’re excited to hear what you think!

Who are Superagents for? Analysts, operators, and anyone doing data-centric work in spreadsheets. We see a tonne of finance people, of course, but also students, researchers and mom & pop shops. Sourcetable's superagents democratize data access and analysis, which is nice because our company’s mission is to make data accessible to everyone.

Why “Superagents”? Because they can plan and orchestrate other task-specific agents to complete your work for you. We have a lot of different AI tools and agents inside Sourcetable, but there’s a whole lot more on the Agentic Web. Superagents are like the conductor that coordinates them all and calls on them when needed. Also, it’s a fun feature name (thanks, Alyssa!)

If you remember the linked-data dream of the semantic web movement, that future is now: all of your business data is available and connected in Sourcetable.

How does it work? Sourcetable is running a python virtual machine under the hood. Everything is sandboxed, and there are hundreds of AI tools and libraries our AI can access. Superagents are also doing code-gen on the fly to solve problems. The closest system we have found is Replit’s sandboxed operating systems. Beyond that Mixtral, ChatGPT and Anthropic offer some limited data connectivity features, except these AI chat services lack the storage, compute, and code execution that Sourcetable and Replit provide. This is all very new.

How is this different to your previous data connectors, etc? We started out using ETL services to sync data and provide a GUI-driven PowerBI like experience in your spreadsheet. This was useful for people who knew SQL and how to write joins to combine fragmented data, but for everyone else (read: practically everyone), this solution just didn’t provide the frictionless, self-serve experience that we wanted.

Our choices were to switch the GTM motion or change the product, so we shelved that reporting suite and focused on our AI spreadsheet and waited for models to catch up with our ambitions. Now that they have, we’re re-launching Sourcetable with our original goal in mind: building a spreadsheet-based operating system for the Agent Web, with fully networked data access for everyone on your team.

AI is the great UX enabler.

Caveats: * We heavily use Postgres, Google Analytics, Stripe and Google Search Console with Superagents. * We haven’t tested every endpoint on the Internet. We find that mainstream, well documented applications work best. * Yes, you can write data back to 3rd party applications and databases. We generally advise against this unless you understand the risks involved in giving AI write-access to your data.

Bonus round: * All data connectors added during this launch week are FREE. (Regular AI messaging limits still apply.)

Product Feedback? [email protected]

I created a canvas-based Markdown table editor, which provides a novel (I think) editing experience. I'd love for people to try it out, or bookmark it for the next time they want to insert a one-off table into a Markdown-formatted comment. My design goals were 1) frictionless user input and 2) fast startup time.

Some good-to-knows:

- Can be used on mobile

- Table data is saved locally to the browser (nothing is sent to a server)

- I only tested this on a MacBook, so let me know if the trackpad calibration feels off

I recently released a new version of my hobby project Vizza. It's a free desktop app of various beautiful visualizations. I've been working on it for several months now and I wanted to share it with y'all.

Paper is a new design tool. We launched into open alpha today. Anyone can now sign up and use Paper.

We started Paper about 1 year ago with the goal to bring more creativity back into design tools. It feels like the existing options are becoming increasingly corporate.

To celebrate to launch, we published a new shader that lets anyone see their logo in Apple's new heat map animation style. There is no sign-up needed at heat.paper.design.

We're always looking for feedback from anyone who uses Sketch, Figma, Photoshop, or Illustrator, about what they most need in their professional design tools.

Have fun with the new shader and please send me anything you make!

Hi everyone, I'm Aman, the creator of DenseWiki.org.

DenseWiki is an experimental deep reading tool.

It aims to amplify human ability to read hard content (research papers, technical articles etc) OUTSIDE our expertise, by rapidly learning new disciplines on the fly.

Here's the key idea (as demonstrated in the video on the website):

When you read something in a new discipline (let's say a paper using AI for biochem, and you nothing about biochem), the challenge is jumping right into an OCEAN of knowledge. You're prone to feel lost and overwhelmed.

DenseWiki's approach is that using the browser extension, if you come across any jargon, it identifies the ONLY few relevant concepts / knowledge you need at that moment, help you quickly become familiar with those few concepts with one click, and let you continue reading.

So as you read, you're able to INCREMENTALLY build your familiarity with the new field and SMOOTHLY expand your knowledge graph, without getting lost — and you're able to engage with the content you want from day 1!

Furthermore, it uses gamification to help you build a consistent deep reading habit.

It also simultaneously builds the world's most cutting-edge knowledge graph — i.e. if you identify a novel concept introduced in a paper that came out only yesterday, you can add it to DenseWiki immediately, making it more advanced than any LLM or blog or web encyclopedia over time.

Looking forward to feedback.

P.S. You'll have to download a browser extension, but if you don't want to sign up, you can log into this test account directly:

Email: [email protected] Password: HACKERNEWS

I built this as a Garage Project over the weekend. https://x402.backpac.xyz/

Still working on refining things, but let me know your thoughts.

Cheers.

This is such a ridiculous hack. A friend with an iphone sent a text to my Android reading:

Like: "I'm thinking of a number"

Google's Messages app interpreted this as:

to "I'm thinking of a number"

I was like, who are you responding to?!

It took a minute of total miscommunication to realize that Messages must be just using the first six characters to decide if something is a "like" command

After struggling to find specific files in my 10,000+ media library, I spent 18 months building RekoSearch. It allows you to search across photos, videos, documents, and audio using natural language queries, such as "dog mountain," or advanced syntax, like `label:couch AND text: "brand name" NOT face:sad`.

Built with Rust (Lambda functions on ARM64 and the entire Backend), Python/React (frontend), AWS services (Rekognition, Textract, Transcribe), and Kubernetes.

• Live demo: https://rekosearch.com/demo

• GitHub repo with architecture docs: https://github.com/Obscurely/RekoSearch-Public (includes in-depth handmade excalidraw diagrams)

• Blog post about my journey and challenges: https://www.adriancrismaruc.com/blog/building-rekosearch-jou...

• Try it yourself (beta): https://rekosearch.com/beta (50 free credits)

I'm open to feedback on architecture, functionality, or any suggestions you have for improvements. Also, I highly recommend trying to build a SaaS app— the knowledge I gained building this far exceeded what I could have learned from smaller desktop projects.

Document is here: https://atsphinx.github.io/qrcode/

Over the past few months, I’ve been building Project Chimera — a prototype AI agent that integrates three reasoning paradigms into a single decision-making system:

- Neuro (LLM) for creative strategy generation

- Symbolic for enforcing business rules and safety constraints

- Causal for predicting the long-term impact of actions

I tested this hybrid agent in a 52-week realistic e-commerce simulation with dynamics such as price elasticity, brand trust, ad ROI, and competitor effects.

Key results:

- Outperformed LLM-only and LLM+Symbolic baselines

- Avoided catastrophic losses by simulating long-term causal impact

- Achieved nearly 2× profit improvement through periodic causal retraining

Resources:

- Live demo: https://project-chimera.streamlit.app/

- GitHub: https://github.com/akarlaraytu/Project-Chimera

I’d appreciate feedback from the HN community — especially on methodology, experiment design, and potential real-world applications beyond e-commerce.

Still a work in progress, but it's usable.

This lets you create "tag points", which are variables with metadata, view and set their state from the web, and create basic automations via the web.

The automations are based on state machines, where each state has a list of "bindings", that set one variable to an expression as long as you're in that state.

Every state also has a variable, so you can enter it by setting it to 1.

There's a (currently unsecured, meant for private WiFi only) WebSockets API, and a Python client.

WiFi Power Saving is supported, power consumption is ~7mA when connected to WiFi, but you have to compile under PlatformIO with the right sdkconfig(There's an example code file).

There's a tiny command shell, which lets you dump the files and then push them to a new device. It is just some regexes, but it pretends to support here docs(They have to start with cat << "---EOF---" > FILENAME ) so that the dumps can be syntax highlighted.

There whole thing is modular, you don't need to use the web based automation features.

One of the modules is MP3/WAV playback, with a background track and an FX track.

Still some bugs, missing features, and especially missing docs, but I thought I'd post it now under the "release early" rule!

Hey HN, I have been working on using the info-gathering and truth-seeking properties of the markets to get more honest reporting on all sort of things. The idea is to have a market game where people can wager a fake currency on what think is true and then, the citizen reviews the crowd-sourced information and makes a judgement about which position has the best case. It is crowd-sourced journalism with a market game. Please let me know what you think and my email is [email protected], if you have any questions.

Hello everyone!

After a few months of working on it in my free time, I just launched my biggest project since I started learning programming in August 2024.

It’s called Stressio, and it’s a tool designed for companies, especially managers. Employees can anonymously log their stress levels, and managers get a heatmap showing which months are the most stressful. This helps teams identify patterns and take action where needed.

There are still some features I want to add, but they’ll come in future updates.

Link: https://stressio.dev

I've just made public a library to perform real time visual saliency detection on videos (but static images are also supported).

This started a couple months ago when, while working on another project, I ended up side-tracking and overkilling as usual. I'm pretty happy with the result and I think it could prove to be a useful piece of software.

It should work on both Linux and macOS, but I'm yet to test Linux cause I only have a mac at hand. Windows may be doable through WSL.

GitHub: https://github.com/big-nacho/dosage Showcase: https://big-nacho.github.io/dosage-docs/showcase.html

I built BPM Finder, a comprehensive audio analysis toolkit that provides three professional-grade tools: BPM detection, musical key identification, and tempo adjustment  all running entirely in your browser.

## What makes it different:

*Privacy-first approach*: Your audio files never leave your device. All processing happens client-side using Web Audio API, rubberband-wasm, and advanced algorithms.

*Three powerful tools in one*: - *BPM Finder*: Dual-algorithm analysis with 99.5% accuracy and confidence scoring - *Key Finder*: Chroma analysis with correlation matrices for precise key detection - *Tempo Change*: Adjust audio speed without pitch changes using professional rubberband algorithm

*Multiple analysis modes*: - Single file analysis with detailed results - Batch processing (up to 50 files simultaneously) - Real-time tap tempo for live BPM detection

## Technical highlights:

- Built with Next.js 15, TypeScript, and Web Workers for intensive processing - Supports MP3, WAV, FLAC, AAC, OGG, M4A (up to 50MB each) - No registration required, completely free to use - Export results to CSV or integrate with DAW/DJ software

## Real-world use cases:

Professional DJs use it for beatmatching in live sets, music producers organize sample libraries by BPM, fitness instructors create tempo-matched workout playlists, and video editors sync background music to cuts.

The project started when I needed reliable BPM detection for my own music production work but couldn't find tools that were both accurate and privacy-respecting. After months of research into audio analysis algorithms and Web Audio API optimization, I built this comprehensive solution.

*Try it*: https://bpm-finder.net/

Would love feedback from the HN community, especially from fellow musicians, developers working with audio processing, or anyone who's tackled similar browser-based analysis challenges.

*Tech stack*: Next.js, TypeScript, Web Audio API, rubberband-wasm, web-audio-beat-detector, Tailwind CSS

Created MangaRenshuu to make Japanese manga reading more effective for learners - click speech bubbles to see extracted text, toggle romaji button if you want to romaji under the extracted text.

Key features: - Pre-processed speech bubble text extraction using OCR - Click bubbles to see Japanese text - Toggle romaji button to see romaji text under extracted text - Currently has some chapters of Yotsuba&! and Takagi-san

Technical process: I used OCR tools to extract and clean text from manga pages offline, then built a web interface that overlays the text on original images.

Started as a personal study tool that I use daily. Made it open-source after realizing it might help other learners.

Would love feedback from other Japanese learners or suggestions for additional manga.

Website: https://www.mangarenshuu.online/ GitHub: https://github.com/MRamazan/MangaRenshuu

## Who is this for?

1) Devs looking for utilities that treat data types generically. For example, `mapValues` takes an array, object, Map or Set and returns a new instance.

2) Devs interested in functional utilities without the jargon. Utilities are named in plain English, there's no currying, and its source is kept simple.

## Other features

- An in-browser playground: https://common-fp.org/try-it

- Supports TypeScript

- 100% test coverage with types tested as well.

- For every utility, I try to explain why it's useful along with a code example.

## A personal note

- I built this as a personal endeavor. I wasn't trying to solve a problem the community expressed, so I understand if people find it odd or unnecessary. Finishing a project is difficult, and I'm mostly happy just to have done that and present what I consider to be a polished product.

## Link to source

- https://github.com/common-fp/common-fp/tree/dev/pkg/common-f...

Thanks for taking a look. ~Phil

* This is a repost since the first time was over a US holiday. Original post here: https://news.ycombinator.com/item?id=45084163

Cjam is lightweight freeware that can process large numbers of MP3 files without quality loss.

Official site: https://cjmapp.net

Also available on MajorGeeks:

https://www.majorgeeks.com/files/details/cjam.html

There’s also a forum where users can share feedback or scripts:

https://forum.cjmapp.net

I'm the founder of eigenarc, an AI powered personalized course generator, that can generate course plan and course content based on your goals, time commitments, what you want to learn or accomplish.

Here are some sample courses: https://eigenarc.com/plan/quantum-computing-mastery-from-beg...

https://eigenarc.com/plan/4-week-senior-software-engineer-in...

Looking for feedback on the product and any feature that would make it more appealing to you.

I’ve released rkik v1.10, a lean Rust tool to inspect NTP servers, compare offsets/RTTs, and optionally perform a one-shot system clock sync on Unix.

Repo: https://github.com/aguacero7/rkik

For those who dont know what it does: - Probe a single NTP server or compare multiple in one shot - Text / Simple / JSON outputs for humans or scripts - Fractional timing controls: sub-second --interval and --timeout (e.g., 0.1, 0.25) - Optional one-shot sync to “server UTC + half RTT” (Unix only) - Dry-run mode to validate the sync workflow without touching the clock

What’s new in v1.10

Clippy integrated in CI with lints as errors (consistent quality) - Sync module public API cleaned up (no more rkik::sync::sync double path) - Sync feature enabled by default in builds/packages - "--interval" and "--timeout" now accept fractional seconds - "--dry-run" for sync to verify flows in CI/scripts

Usage

Installing it is very simple using cargo: cargoo install rkik

Single server probe: rkik time.google.com

Compare multiple servers (10 requests, 200 ms apart), simple text: rkik -C time.google.com time.cloudflare.com --count 10 --interval 0.2 --short

JSON output (pretty-printed): rkik pool.ntp.org -jp

One-shot sync (Unix; needs root or CAP_SYS_TIME): rkik --server time.google.com --sync

Notes / caveats - Sync requires privileges (root or CAP_SYS_TIME). Dry-run works without them. - If a time daemon (chronyd/systemd-timesyncd) is active, it may readjust your clock afterward; that’s expected.

Roadmap / feedback I’m aiming to make it a PTP / NTS protocols Diag tool, I want rkik to become a real Time servers monitoring and analysis toolbox. Suggestions on output ergonomics, useful stats for CI/observability, or packaging for your distro or even github contributions are very welcome.

Happy to answer questions and take feedback!

A Terraform-based starter kit for automating Azure infrastructure with GitHub Actions.

UnmetNeeds is a crowdsourced platform by Virock where people share ideas for products or solutions they wish existed. Others can vote, pledge payment, or propose answers—turning collective wants into future innovation.

I’ve been working on an open-source CLI tool called Dochia for a while now.

The reason for building it: in microservices architectures, testing is repetitive: every service exposes APIs, and much of the testing process will share a similar set of test cases: negative inputs, boundary cases, encoding quirks, invisible characters. I wanted a tool that could automate this shared layer of testing instead of doing it manually across dozens of services.

So what Dochia does: reads your OpenAPI spec; generates thousands of smart, edge-case payloads automatically and produces detailed reports with reproducible cases.

I've been use it for any microservice I've built as to establish a foundation in place.

Github: https://github.com/dochia-dev/dochia-cli and website: https://dochia.dev

It's open-source, and I'd love your feedback. Contributions, issues, and edge-case stories are very welcome.

I started this project from scratch 3 years ago before cs2 was released.

Now game has all csgo features implemented notably: multiplayer, weapon shooting, grenade throwing, package planting, round end win conditions, buy menu, scoreboard, radar, economy, recoil patterns, sniper rifle, proper hitboxes with backtrack support, inventory managment and legendary dust2 map

what is missing is better audio and visuals as I am not really good at this category

if you wanna try game yourself you can download it here https://solcloud.itch.io/counter-strike-football#download or build from source code

any feedback is welcome, thank you for reading

GitHub: https://github.com/Ferki-git-creator/link-cli-tool