2025年12月23日 的 Show HN

Hi HN

I built CineCLI — a cross-platform terminal app to browse movies, view details, and open torrents directly in your system torrent client.

Features: - Search movies from the terminal - Rich UI with ratings, runtime, genres - Interactive & non-interactive modes - Magnet handling via system default client - Linux/macOS/Windows support - No ads, no tracking

GitHub: https://github.com/eyeblech/cinecli PyPI: https://pypi.org/project/cinecli/

Would love feedback from terminal + Python folks

I made a web game inspired by Geoguessr and Wordle, where you get shown a 3D model of a random human protein each day, and you have to triangulate its gene name using similarity clues.

My background is in wet lab molecular biology and I intend this game to be engaging mostly to other biologists. But if you're outside the field, I'm interested to know if you can still solve it with browser use LLMs, and if you learned something interesting doing so. Let me know what you think.

I made it with Claude over the last 2 months. My coding experience is limited to basic python data analysis and figure making. I've seen people online asking, "Now that we have coding AI, why isn't there a deluge of awesome AI-generated apps made by non-coders?" - if this sounds like you, check out Geneguessr to understand what a web app by a non-coder looks like.

I might write more about the process if there's a demand, but what really unlocked the project for Claude was Linear MCP, where it could put each individual issue on a shared Kanban board. This, and Playwright MCP for testing on live site, were the two workhorses that got me through this. For bugs Claude couldn't one-shot, Linear was great for consolidating issue information so that I could dump it into ChatGPT Codex - it would usually think for like half an hour, output very confusing explanations, but the bug was gone.

Game is free, no log-in required, sorry if you run into any mobile bugs - didn't test it much there.

https://geneguessr.brinedew.bio/

Hey HN, I'm Andres. I've been building Kapso as a solo founder, and just crossed 4,000 developers, all organic.

WhatsApp has 3B+ users and 98% open rates. You'd expect developers to be building tons of stuff on it, especially when the US is the fastest-growing market in WhatsApp usage.

But it’s not happening… And I'd bet it's because the DX is painful.

Every team needs to build the same features again and again. Meta fires webhooks for everything. There's valuable data in there for debugging, but no way to make sense of it without building your own tooling.

That’s why I built Kapso. What you get:

- Working WhatsApp API + inbox in 2 minutes, not days - Full observability: every webhook parsed, every message tracked, actual debugging tools - Multi-tenant platform: generate a setup link, customer connects their Meta account, done - Workflow builder for deterministic automations and AI Agents - WhatsApp Flows: build mini apps inside WhatsApp using AI + serverless functions - Docs that work for humans and LLMs

We're up to 95% cheaper than Twilio, with a generous free tier (2,000 messages per month).

We also open source several tools: a TypeScript client for the WhatsApp Cloud API, a reference WhatsApp Inbox implementation, and a voice AI agent for WhatsApp.

GitHub: https://github.com/gokapso/

Happy to answer questions!

https://kapso.ai/

Claude Code added a /stats command. The stats are cached in $HOME/.claude, so I made a fun program with Bun + WASM that pulls the stats (nonsensitive, nonidentifiable), posts them to a database, and lets you see where your usage lies.

The code is here[^1] if anyone's interested in the rendering code or wants to make sure I'm not exfiltrating their credit card.

[^1]: https://github.com/tspader/claude-wrapped/

A colleague and I made a kids' picture book that introduces authorization concepts.

We work at AuthZed and explain these concepts regularly. We thought it'd be fun to put them together in a format accessible and appealing to kids and grownups alike. It would also be helpful when explaining what we do for work and make a unique gift for our families.

The goal was a fun story first and foremost. We aimed to present concepts accessibly but made conscious decisions to simplify, knowing we couldn't be comprehensive in a picture book format. We also wanted visually appealing illustrations, so we built a custom tool to streamline exploring ideas with AI. It does reference-weighted image generation (upload references, weight which ones matter most), git-like branching for asset organization, and feedback loops that improve subsequent generations. It was built with Claude Code. Here's a screenshot: https://content.authzed.com/hubfs/screenshot-design-asset-ma...

We'd love feedback on where we chose to simplify. Did we get the tradeoffs right or did we oversimplify? And lastly, did you enjoy the story?

You can read the book online: https://authzed.com/resources/dibs-and-the-magic-library

Hi HN!

I built Openinary because Cloudinary and Uploadcare lock your images and charge per request.

Openinary lets you self-host a full image pipeline: transform, optimize, and cache images on your infra; S3, Cloudflare R2, or any S3-compatible storage.

It’s the only self-hosted Cloudinary-like tool handling both transformations and delivery with a simple URL API (/t/w_800,h_800,f_avif/sample.jpg).

Built with Node.js, Docker-ready.

GitHub: https://github.com/openinary/openinary

Feedback welcome; especially from Cloudinary users wanting the same UX but on their own infra!

Over the past couple of months, I've been working on gonc. It started as a cross-platform netcat alternative written in Go, but I realized it would be much more powerful with a -p2p flag to automatically punch through NATs and firewalls.

It is now working smoothly. To establish a P2P netcat-like chat connection, you simply run the same command on both sides:

gonc -p2p <passphrase>

Building on this, I implemented a dual (bidirectional) reverse proxy. This allows both peers to access each other's networks simultaneously.

Peer A:

gonc -p2p <passphrase> -linkagent

Peer B:

gonc -p2p <passphrase> -link 1080;1080

This opens port 1080 on both ends as a SOCKS5 & HTTP proxy to tunnel traffic to the other side.

How it differs from frp/rathole: Unlike existing solutions that require you to deploy a self-hosted Rendezvous/Signaling Server, gonc uses public STUN and MQTT servers for signaling.

For cases involving Symmetric NATs where a relay is unavoidable, gonc doesn't need a proprietary relay server—it can simply use any standard SOCKS5 server as a relay.

Peer B: (only one side needs to specify -x)

gonc -x socks5ip:port -p2p <passphrase> -link 1080;1080

I'd love to hear your feedback or suggestions!

Successful entrepreneurs always say that the most profitable tools are the ones that help you solve the issues you’re facing. The problem is, I apparently have no issues. So instead, I built a PoC of a minimalistic ideas directory focused on issues others are facing. Feedback is welcome

Hey HN! I'm Baha, creator of Mysti.

The problem: I pay for Claude Pro, ChatGPT Plus, and Gemini but only one could help at a time. On tricky architecture decisions, I wanted a second opinion.

The solution: Mysti lets you pick any two AI agents (Claude Code, Codex, Gemini) to collaborate. They each analyze your request, debate approaches, then synthesize the best solution.

Your prompt → Agent 1 analyzes → Agent 2 analyzes → Discussion → Synthesized solution

Why this matters: each model has different training and blind spots. Two perspectives catch edge cases one would miss. It's like pair programming with two senior devs who actually discuss before answering.

What you get: * Use your existing subscriptions (no new accounts, just your CLI tools) * 16 personas (Architect, Debugger, Security Expert, etc) * Full permission control from read-only to autonomous * Unified context when switching agents

Tech: TypeScript, VS Code Extension API, shells out to claude-code/codex-cli/gemini-cli

License: BSL 1.1, free for personal and educational use, converts to MIT in 2030 (would love input on this, does it make sense to just go MIT?)

GitHub: https://github.com/DeepMyst/Mysti

Would love feedback on the brainstorm mode. Is multi-agent collaboration actually useful or am I just solving my own niche problem?

Hi HN — I built Ayder because Kafka in 2025 feels like starting a car with a hand crank.

# That's it. You're streaming. curl -X POST localhost:1109/broker/topics/orders/produce -d '{"item":"widget"}'

Single binary. Zero dependencies. HTTP-native (not a proxy). Built in C with libuv.

Performance (wrk2, real network, 3-node Raft, sync-majority writes):

  Throughput:     50K msg/s sustained
  P99:            3.46ms
  P99.9:          12.94ms
  Server P99.999: 1.22ms (HTTP isn't the bottleneck)

Recovery time — the number nobody talks about:

  Kafka after unclean shutdown: 2+ hours (real production reports)
  Ayder after SIGKILL: 40-50 seconds with 8 million offsets

Tested repeatedly. Zero data loss. Follower replays AOF, asks leader for missing offsets, catches up automatically. No manual intervention.

What you get: - Append-only log with consumer groups + committed offsets - Raft consensus (3/5/7 nodes) with mTLS - KV store with CAS + TTL - Stream processing: filter, group_by, windowed joins (including cross-format Avro+Proto) - Idempotent produce, retention policies, Prometheus metrics

I'm a solo founder from Kazakhstan, self-taught systems programmer. Spent 8 months building this. Looking for feedback on the API, what's missing, and early design partners.

GitHub: https://github.com/A1darbek/ayder

Hi, I wanted to create an AI powered team, where agents with different skills and tools could collaborate with each other on complex tasks.

But the current solutions felt too developer oriented. I wanted something that felt closer to how humans actually work. So I decided to create Starships.ai, where working with agents feels like talking to a remote employee on Slack.

The goal is to create an organization that is primarily run by agents, with humans reviewing critical decisions.

I would love to hear what you think, and would love for you to give it a try.

Thanks

https://starships.ai

I built an extension that gives developers cheaper access to AI models in a Kilo Code fork.

Free tier: 200 requests/day to open source models (DeepSeek, Qwen, Minimax) Paid tier ($20/month): $45 in credits for Claude/GPT-5/Gemini-3 + 1000 daily open source requests

Works in VS Code, Cursor, and Windsurf. Your data isn't used for training.

Just launched today, still rough but functional. Would appreciate feedback.

agentica.genlabs.dev Download extension from: https://open-vsx.org/extension/agentica/agentica

Hey HN, I'm a SysAdmin turned Dev.

Leaving your mining rigs unmonitored for the holidays is a risk, so I built a tool to automate it.

I got tired of the delayed/fragmented data from ISOs like PJM and ERCOT, so I built a normalized API to track grid stress in real-time.

The Stack:

Scrapers: Python 3.11 (Pandas/Requests) handling the messy ISO formats. Compute: Azure Functions (Consumption Plan) to keep costs near zero. Storage: Azure Data Lake Gen2 (Parquet) for historical retention.

I also wrote a simple Python client 'kill-switch' for Bitcoin miners to automate curtailment during price spikes.

Repo is here: https://github.com/Norris-Eng/gridwatch-kill-switch API is here: https://rapidapi.com/cnorris1316/api/gridwatch-us-telemetry

Happy to answer questions about the scraping logic or Azure costs!

Hi HN,

I built VeriMed while working on an AI-powered telemedicine platform. I quickly realised that verifying whether a healthcare provider is legitimately licensed is surprisingly hard.

Every country has a different registry, a different API format (REST, SOAP, FHIR), and different access requirements. I couldn't find a unified solution, so I built one.

VeriMed: - Connects to 5 national medical registries: - USA (NPI) - France (RPPS) - UAE (DHA) - Netherlands (BIG) - Israel (MOH) - Falls back to AI document verification when registries aren't available - Uses fuzzy name matching to handle variations (e.g., "Greg" vs "Gregory") - Production-ready with Docker, Kubernetes manifests, and health checks - MIT licensed, free to self-host - Enterprise extension available: - SSO (SAML 2.0 & OIDC) - RBAC (Admin, Reviewer, Viewer) - Audit Dashboard & CSV Export - Bulk CSV Import Wrapper

Tech stack: NestJS, TypeORM, OpenAI (optional), Fuse.js

GitHub: https://github.com/daretechie/verimed

I'd love feedback on the architecture or suggestions for additional country adapters to prioritise.

Hi HN,

Like many here, I started trading the "Wheel" strategy (selling puts/covered calls) to generate income. I quickly realized that tracking my true cost basis was a nightmare.

My brokerage shows the stock's purchase price, but it doesn't "know" that I've collected $400 in premiums over the last three months. I ended up with a massive spreadsheet that I had to update manually every Friday just to see if I was actually profitable.

I built PremiumFlow to automate this. It treats the option premium and the underlying stock as a single financial unit.

What’s inside:

Automatic Basis Adjustment: If you sell a call against a LEAP or stock, your "break-even" price updates instantly.

Assignment Risk: It monitors how close your short puts are to being ITM.

No Spreadsheet Mess: All the math (annualized returns, DTE, etc.) is handled in the dashboard.

I’d love for you to kick the tires and let me know what you think. I'm especially interested in feedback on the UI and any edge cases in your own options strategies that I might have missed!

I built a drag-and-drop map playground to compare the true sizes of countries and visualize Mercator distortion, just for fun and sharing. Open source + easy to replace the GeoJSON with your own data.

Demo: https://www.runcell.dev/tool/true-size-map

Source: https://github.com/ObservedObserver/world-map-reality

I built this because I consume a lot of technical lectures and long-form podcasts on YouTube, but I found myself wasting hours scrubbing through videos just to find specific citations or concepts.

Recapio is a tool that extracts the transcript and generates structured summaries for videos (and web articles). It’s not trying to replace watching content, but rather to act as a 'Ctrl+F' for video context.

One technical challenge I faced: Dealing with auto-generated YouTube captions vs. forced captions was messy. I had to build a parser that normalizes the timestamps so that when you click a summary point, it actually seeks to the correct frame, even if the caption timing is drifting.

It has a free tier that should cover most casual usage. I’d love your feedback on the extraction quality

Hi HN,

Like many people, I was frustrated that the released Epstein/Maxwell court documents were mostly scanned images (PDFs) with no text layer. This made them impossible to Ctrl+F or analyze programmatically.

I built a pipeline to fix this using Python, Tesseract, and OpenSearch.

The Site: https://epsteinfilez.com

The Stack:

Ingestion: Python workers using ocrmypdf (Tesseract) to perform parallel OCR on raw files.

Search: OpenSearch for indexing the extracted text.

Frontend: Next.js (SSR) for the UI.

Infrastructure: Self-hosted Docker swarm.

Features:

Sub-second full-text search across ~15,000 pages.

Highlights search terms directly on the PDF page.

Deep linking to specific pages/documents.

This is a transparency tool, not a political one. I wanted to make the raw primary sources accessible to researchers and journalists.

Feedback on the search relevance or indexing pipeline is welcome!

Hello, all. Luxury Yacht is a desktop app for managing Kubernetes clusters that I've been working on for the past few months. It's available for macOS, Windows, and Linux. It's built with Wails v2 (https://wails.io), so huge thanks to Lea Anthony for that awesome project. Can't wait for Wails v3.

This originally started as a personal project that I didn't intend to release. I know there are a number of other good apps in this space, but none of them work quite the way I want them to, so I decided to build one. Along the way it got good enough that I thought others might enjoy using it.

Luxury Yacht is FOSS, and I have no intention of ever charging money for it. It's been a labor of love, a great learning opportunity, and an attempt to try to give something back to the FOSS community that has given me so much.

A word of caution, in case you care about this sort of thing -- I wrote this with AI assistance. I am, first and foremost, a platform/infrastructure engineer, not a software developer. I've learned enough via tinkering and osmosis over the years to have a decent idea of how to write software, but it's not my main gig. I did my best to try to make sure the code isn't garbage, but you might look at it and disagree. I have no doubt that there's a lot of room for improvement, but it works, I think it has a pretty good feature set, and I'll keep iterating on it to make it better.

If you want to get a sense of what it can do without downloading and installing it, read the primer: https://github.com/luxury-yacht/app/blob/main/docs/primer.md

Or, head to the Releases page to download the latest release: https://github.com/luxury-yacht/app/releases

Oh, a quick note about the name. I wanted something that was fun and invoked the nautical theme of Kubernetes, but I didn't want yet another "K" name. A conversation with a friend led me to the name "Luxury Yacht", and I warmed up to it pretty quickly. It's goofy but I like it. Plus, it has a Monty Python connection, which makes me happy.

Healthy Base automates PostgreSQL backups so you never have to worry about cron jobs failing silently or lost data.

Features: - Automated scheduled backups (or manual with one click) - Encrypted cloud storage with versioning - Email alerts for backup status and errors - Mount backups to inspect data without full restore - Free tier available with 3 backups per project

Stop babysitting pg_dump scripts. Start sleeping easy.

Hi HN,

A few months ago I posted JotChain here — a work log tool that tried to do too much. The feedback was fair: the value wasn’t obvious.

I rebuilt it with a much narrower focus: performance reviews.

The problem

We recently started doing performance reviews, and I had to review three developers. I remembered some highlights and issues from the past year, but a lot of context was missing — especially small but important things like soft-skill feedback or moments where someone quietly helped unblock the team.

Those things never make it into a 1:1 doc, and by review time they’re gone.

I wanted one place where I could quickly jot things down as they happen, without worrying about structure — and only structure it when it matters.

The approach

Write short notes as work happens (30 seconds, no format)

Add light context (project, teammates)

Pick a date range and generate a clean: self-review, peer/person review, accomplishments summary.

No meeting recordings, no standup bots — just a lightweight log that turns into review-ready output.

It’s early and adoption is slow, so I’m mainly looking for feedback:

- Does this solve a real problem for you? - How do you handle performance reviews today? - Is “performance reviews” a strong enough wedge?

Live here: https://jotchain.com

Happy to answer questions and very open to criticism.