2025년 12월 8일의 Show HN

Hey HN! Like most here regular meetings have always been a big part of my work.

Over the years I've learned the value of active note taking in these meetings. Meaning: not minutes, not transcriptions or AI summaries, but me using my brain to actively pull out the key points in short form bullet-like notes, as the meeting is going on, as I'm talking and listening (and probably typing with one hand). This could be agenda points to cover, any interesting sidebars raised, insights gotten to in a discussion, actions agreed to (and a way to track whether they got done next time!).

It's both useful just to track what's going on in all these different meetings week to week (at one point I was doing about a dozen 1-1s per week, and it just becomes impossible to hold it in RAM) but also really valuable over time when you can look back and see the full history of a particular meeting, what was discussed when, how themes and structure are changing, is the meetings effective, etc.

Anyway, I've tried a bunch of different tools for taking these notes over the years. All the obvious ones you've probably used too. And I've always just been not quite satisfied with the experience. They work, obviously (it's just text based notes at the end of the day) but nothing is first-class for this usecase.

So, I decided to build the tool I've always felt I want to use, specifically for regular 1-1s and other types of regular meetings. I've been using it myself and with friends for a while already now, and I think it's got to that point where I actually prefer to reach for it over other general purpose note taking tools now, and I want to share it more widely.

There's a free tier so you can use it right away, in fact without even signing up.

If you've also been wanting a better system to manage your notes for regular meetings, give it a go and let me know what you think!

Hi!

I got tired of setting up tools I can't explain to a team in a few words like sops or git-crypt, just to store few files with environment variables or secrets, so I built lockenv as a simple alternative.

It's basically a password-protected vault file you commit to git. No gpg keys, no cloud, just lockenv init, set a password, and lock/unlock the secrets.

This tool integrates with OS keyring, so you're not typing passwords constantly. Should work on Mac/Linux/Windows, but I tested it only on linux so far.

I am not trying to replace any mature / robust solution, just making small tool for simple cases, where I want to stop sharing secrets via slack.

Feel free to try, thank you!

Hello Everyone! We built SQLFlow as a lightweight stream processing engine.

We leverage DuckDB as the stream processing engine, which gives SQLFlow the ability to process 10's of thousands of messages a second using ~250MiB of memory!

DuckDB also supports a rich ecosystem of sinks and connectors!

https://sql-flow.com/docs/category/tutorials/

https://github.com/turbolytics/sql-flow

We were tired of running JVM's for simple stream processing, and also of bespoke one off stream processors

I would love your feedback, criticisms and/or experiences!

Thank you

Hi HN,

I built WhatHappened (whathappened.tech) because I have a love/hate relationship with this site. I love the content, but the "wall of text" UI gives me FOMO. I was spending too much time clicking into vague titles ("Project X") or wading through flame wars just to find technical insights.

I built this tool to act as a filter. It generates a card for the top daily posts with a few specific features to cut the noise:

1. AI Summaries: It generates a technical TL;DR (3 bullet points) and an ELI5 version for every post.

2. The Heat Meter: I analyze the comment section to visualize the distribution: Constructive vs. Technical vs. Flame War. If a thread is 90% Flame War, I know to skip it (or grab popcorn).

3. Contrarian Detection: To break the echo chamber, the AI specifically hunts for the most upvoted disagreement or critique in the comments and pins it to the card.

4. Mobile-First PWA: I mostly read HN on my phone, so I designed this as a PWA. It supports swipe gestures and installs to the home screen without an app store.

Stack: Next.js, Gemini, Supabase.

It currently supports English and Chinese. Any feedback will be appreciated! My original X post: https://x.com/marsw42/status/1997087957556318663, please share if you like it or find it helpful! :D

Thanks!

I had an issue where I would journal stuff, and then never look at what I wrote. So I thought it'll be cool to schedule something that will get sent to you at a later time (like a time capsule). Also, was inspired by futureme, where you can send yourself letters that'll arrive in the future.

A small tool that parses C declarations and outputs a simple visual representation at each stage, as it encounters arrays, pointers or functions.

The program uses a table-driven lexer and a hand-written, shift-reduce parser. No external dependencies apart from the standard library.

LinkedQL is a new SQL client that supports live queries over any Postgres, MySQL, and MariaDB database. You get result sets that self-update differentially as rows change in your database – via inserts, updates, deletes. Works with no extra tooling/ORM layer or GraphQL servers. You opt into live mode simply with a flag: client.query('SELECT ...', { live: true }). More at: https://linked-ql.netlify.app/capabilities/live-queries

LinkedQL is written in JavaScript and runs in both client and server environments.

GitHub + docs: https://github.com/linked-db/linked-ql

Demo examples included.

I’d love feedback: • Anything confusing? • Anything seems useful or dangerous? • Anything else that'd make you consider LinkedQL for production?

Thanks for taking a look — happy to answer any questions.

Hi HN, I have been building voice agents for sometime now. I was earlier automating parts of visa processing, and we needed real-time, multilingual voice calling.

I assumed the hard work was just wiring LiveKit/Pipecat + STT/TTS + an LLM. It wasn’t.

Even with solid OSS (Pipecat/LiveKit), we still had to do a lot of plumbing- variable extraction, tracing, testing etc and any workflow changes required constant redeploys.

We eventually realized we’d spent more time building infrastructure than building the actual agents. Everything felt custom. We hit every possible pain with Pipecat and VAPI style systems.

So we built Dograh - a fully open-source voice agent framework that includes all the boring, painful pieces by default.

What’s different:

- Pipecat-based engine, but forked - custom event model, and concurrency fixes

- One-click start template generated by an LLM Agent for a quick get start template for any use case

- Drag-and-drop visual agent builder for quick iteration (the thing we wished existed earlier)

- Variable extraction layer (name/order/date/etc.) baked into the LLM loop

- Built in Telephony integration (Twilio/ Vonage/ Vobiz/ Cloudonix)

- Multilingual support end-to-end

- Select any LLM TTS STT (add their credits, if any)

- AI-to-AI call testing: automatically stress-test an agent before shipping (still a work in progress- so patchy as of now)

- Fully Open Source

It's built and maintained by YC alumni / exit founders who got tired of rebuilding the same plumbing.

Why we open-sourced it: We kept feeling that the space was drifting toward closed SaaS abstractions (VAPI, Retell). Those are good for demos, but once you need data controls, privacy or self/offline deployment, you end up stuck. We wanted a stack where you can see every part, fork it, self-host it, and patch it as needed.

Try it:

- Repo: https://github.com/dograh-hq/dograh

This spins up a basic multilingual agent with everything pre-wired.

Who this is for:

- If you are looking for self hosting a Vapi like platform for Data Privacy etc.

- Anyone trying to build production-grade voice agents without reinventing audio plumbing.

- If you’ve tried to glue STT→LLM→TTS manually, you probably know the exact pain this is built for

Happy to answer technical questions, show the architecture, or hear how we can improve the product.

I was going through my app subscriptions and realized I was paying $15 for a pretty good workout app, which seemed a bit high to me.

As a software engineer who is also well versed in claude code, I realized that I could likely vibecode a very similar app, or even build something more to my liking. I challenged my self to build something roughly equivalent this afternoon.

Workflow was: start with a detailed spec from Claude code describing many of the features common in workout apps. Then paste this into lovable to have it build out the initial mvp.

Once that was built, I used claude code extensively to modify the app until it was usable, including adding an import from the costly premium app.

While there are bugs, I think I might use this app. And it is insane that we are in a place where I can build this on my phone during an afternoon. In a few years, the economics of apps is going to be different, at least for folks willing to work a little bit.

In theory this project will save me over $190 a year.

Hey HN! I'm open-sourcing DataKit today.

GitHub: https://github.com/datakitpage/datakit Live demo: https://datakit.page

DataKit is a browser-based data analysis platform that processes multi-gigabyte files (CSV, Parquet, JSON, Excel) entirely client-side using DuckDB-WASM. Your data never leaves your browser.

What it does: • Process large files (tested up to 20GB) without any server • Full SQL interface powered by DuckDB compiled to WebAssembly • Python notebooks via Pyodide for data science workflows • Connect to remote sources (PostgreSQL, MotherDuck, S3) with optional proxy • AI assistant that only sees column schemas, not actual data

I was done with having to choose between cloud tools and heavy local installations. I wanted something that just works in a browser tab but has real power.

It's AGPL licensed with commercial licenses available for enterprises.

I've been building this solo as a side project for the past few months. Would love your feedback on: - Performance bottlenecks you encounter - Features you'd need for your workflows - The architecture decisions (all client-side vs hybrid)

Hey everyone I’ve been on my FIRE journey for a while and got tired of juggling spreadsheets, brokers, and bank apps — so I built WealthYogi, a privacy-first net worth tracker focused on clarity and peace of mind. Why Like many FIRE folks, I was juggling spreadsheets, bank apps, and broker dashboards — but never had one clear, connected view of my true net worth. Most apps required logins or shared data with third parties — not ideal if you care about privacy. So I built WealthYogi to be: Offline-first & private — all data stays 100% on your device Simple — focus purely on your wealth trajectory, not budgeting noise Multi-currency — 23 currencies, supporting GBP, USD, EUR, INR and more What it does now * Tracks your net worth and portfolio value in real time * Categorises assets (liquid, semi-liquid, illiquid) and liabilities (loans, mortgages, etc.) * Multi-currency support (GBP, USD, EUR, INR and more) * Privacy-first: all data stays 100% on your device * 10+ Financial Health Indicators and Personalised Finance Health Score and Suggestions to improve * Minimal, distraction-free design focused purely on your wealth trajectory Planned features (already in development) Real-time account sync Automatic FX updates Import/Export support More currency account types Debt tracking Net worth forecasting Pricing Free Trial for 3 days. One time deal currently running till 10th December. Monthly and Yearly Subscriptions available.

Would love your feedback 1. Try the app and share honest feedback — what works, what feels clunky 2. Tell us what features you’d love to see next (especially FIRE-specific ideas!) 3. Share how you currently track your net worth — spreadsheet, app, or otherwise

Here’s the link again: WealthYogi on the App Store (https://apps.apple.com/us/app/wealthyogi/id6753881658) WealthYogi on the Android (https://play.google.com/store/apps/details?id=com.dyogi.weal...) Demo (https://youtu.be/KUiPEQiLyLY)

I am building this for the FIRE and personal finance enthusiasts, and your feedback genuinely guides our roadmap. — The WealthYogi Team [email protected]

I built a small weekend project to help track RAM prices, since DDR3/DDR4/DDR5 costs have suddenly jumped recently and I was struggling to find good deals for my NAS build.

RamScout scans eBay (UK/US) and ranks RAM listings by price per GB, with filters for type, capacity, speed, condition, etc. It’s a simple MVP — no frills, no accounts, no ads — just a fast way to spot unusually cheap listings.

Would appreciate any feedback, especially on performance, UI, and whether expanding to more regions/vendors would be useful. Thanks!

Me and my friend built Leetwrap — a "Spotify Wrapped" style yearly summary for LeetCode users.

It shows your stats, problem streaks, ranking distribution, and a few fun visuals based on your LeetCode activity.

Would love feedback from the community. Still improving the logic and visuals, so suggestions are welcome!

Hi HN!

I built an open source macOS app that allows you to unlock your SSH and GPG key passphrases with Touch ID. You can also store secrets and inject them into your dev environment using the CLI, and store `age` encryption keys in the Secure Enclave.

I started working on this because I was setting up a new computer, and I didn't like how janky the Mac GPG pinentry tool felt.

Secrets management came later because I also wanted an alternative to 1Password's secrets injection, which only works while online for some reason. This allows me to store `axo://...` URLs in my config files, which get dynamically populated with `ap inject`. The vault spec is inspired by SOPs.

The reason this is an app and not just a CLI tool is because it integrates directly with Apple's Security framework and needs to be codesigned and notarized - an interesting (annoying) problem I should write about

Looking to the future, I'm planning to add support for syncing secret vaults with git, SOPs-backed vaults, syncing public keys to Github, GPG key management (I keep forgetting how to renew my keys), and maybe even implement the ssh-agent protocol so I can store my SSH keys in the app. It's a long list but it will solve many paper cuts of mine.

Would love some early feedback, happy to answer any questions.

Hey HN,

I built this because I got tired of "silent failures" in traditional PDF table extraction tools.

In my day job working with financial and legal documents, tools like Camelot or Tabula often return data that looks plausible but has shifted columns or missing decimal points. In regulated environments, you can't afford to guess.

I built a pipeline that treats extraction as a hypothesis to be verified:

1. *Extraction:* Uses IBM’s Docling to parse the layout and get the structure (Markdown).

2. *Visual Verification:* Captures a screenshot of the specific table region from the PDF.

3. *Validation:* Feeds both the Markdown and the Screenshot into a local Vision LLM (Llama 3.2 via Ollama).

4. *Scoring:* The LLM compares pixel truth vs. extracted text and outputs a confidence score + audit trail.

The trade-off is speed (it takes ~5s per table) vs. confidence. It's designed to run 100% locally for privacy-critical documents.

Repo is here: https://github.com/2dogsandanerd/validated-table-extractor

Would love to hear how you handle data integrity in RAG pipelines!

Hi HN,

Edge.mq makes it very easy to ship data from the edge to S3.

EdgeMQ is a managed HTTP to S3 edge ingest layer that takes events from services, devices, and partners on the public internet and lands them durably in your S3 bucket, ready for tools like Snowflake, Databricks, ClickHouse, DuckDB, and feature pipelines.

Design focus on simplicity, performance and security.

I’ve been working on a VS Code extension called Nogic that indexes your project and turns it into an interactive graph of files, symbols, imports, calls, and references. The goal is simple: make it much easier for developers to understand how a codebase fits together, especially now that AI is generating more code than ever.

Hi everyone,

I really, really enjoy playing Wordle, but my limited vocabulary often leaves me stuck. So I ended up building a small helper to give myself a hand while playing.

It's not meant to "cheat" at Wordle — the idea is just to help you practice more often and get better at guessing words. I kept it lightweight and interactive so it still feels like you’re actually playing and learning as you go.

A few things I found interesting while building it:

1. The candidate list updates live as you type—no precomputed tables 2. The scoring model combines basic frequency weights and positional likelihood 3. Keeping the UI minimal made the tool feel more like a helper than a solver 4. I optimized filtering so it stays responsive even on low-end mobile devices 5. I added a dark theme mostly because I play late at night

Some technical notes for anyone curious:

1. All filtering and scoring happens client-side 2. The word list is cleaned up to avoid overly obscure entries, but I'm open to suggestions 3. I've been experimenting with different weighting strategies and would love feedback

If you like playing wordle games, I'd love to hear your thoughts. Feedback, ideas, or nitpicks are all welcome. Glad to answer any questions!

Hey HN,

I built a static site, no login, no database, no apps, with weekly challenges designed to break dopamine loops, disrupt routines, and help you appreciate what you already have.

Every week, a new challenge drops: no moving frames, candlelight evenings, corded phones, sleeping on the floor, and more.

Here is the contract I generated for myself this week (Vol 50):

I am locking in for WaldenWeek Vol 50: Corded Phone. The Rule: Phone stays plugged in at home. 24/7. No exceptions. The Commitment: If I fail, I will Venmo $10 to the first friend who calls me out.

Who is brave enough to join me?

Rules and challenges are available at: https://waldenweek.com

Hi HN,

Not too long ago, I learned that longterm Linux kernels contain a lot of known and unpatched CVEs, in hindsight this makes sense, I was just a bit surprised at the time.

Long story short, I've made a website that lists all the known CVEs for all kernel versions since 2.6.11.

It makes use of the output of tooling provided by the kernel maintainers, so the list of CVEs should be accurate.

The system has an API that's documented under the 'Docs' section.

Let me know if you find this useful and how I can improve it or add features.

Cheers

Hi HN, I built ExtremeFitness to help personal trainers manage workout and diet plans without relying on PDFs, WhatsApp, or Google Sheets.

It allows trainers to: - Create personalized workout plans - Add diet plans - Use a 700+ exercise library - Let clients access everything in one place - Use AI to assist in plan creation

This is still early-stage and I’m actively improving it. I would really appreciate honest feedback on the product, UX, and feature direction.

Happy to answer any technical questions.

I kept writing the same brittle shell one-liners in Dockerfiles to check if binaries exist, env vars are set, and config files are valid. The error messages were always terrible and the syntax inconsistent.

Preflight replaces all that with a single static binary. It handles commands (with version constraints), env vars, files, TCP/HTTP endpoints, checksums, git state, and system resources. Works in FROM scratch images since it has zero dependencies.

Happy to hear what validation patterns I'm missing.

Axis is an experiment in designing a small, deterministic, semantics-first logic language with an AI collaborator. The goal is to explore whether a minimal, rigorously defined semantic layer can help AI systems generate safer and more consistent code across different host languages.

The repo contains the draft whitepaper, early semantics. This is very early work — I’m sharing it now to get feedback from people working in PL theory, compilers, formal methods, and AI tooling.

Repo: https://github.com/axis-foundation/axis-research

Whitepaper (PDF): https://github.com/axis-foundation/axis-research/blob/main/p...

I’m particularly interested in where this overlaps with existing research, where the ideas may be flawed, and whether the overall direction seems useful or misguided. All feedback welcome.