Hacker News

Article URL: https://platypi.empla.io

Comments URL: https://news.ycombinator.com/item?id=47160362

Points: 1

# Comments: 1

Article URL: https://bloomfilter.xyz/

Comments URL: https://news.ycombinator.com/item?id=47160347

Points: 1

# Comments: 0

Article URL: https://elpais.com/america/lideresas-de-latinoamerica/2026-02-25/genero-racismo-y-xenofobia-asi-son-los-sesgos-de-la-inteligencia-artificial-en-latinoamerica.html

Comments URL: https://news.ycombinator.com/item?id=47160345

Points: 1

# Comments: 0

Scan any site and generate WebMCP tool definitions for AI agents. Open source CLI. Deploy with one script tag, test in an agent playground, and A/B test your definitions to find what agents prefer. All at agents.keak.com.

Comments URL: https://news.ycombinator.com/item?id=47160342

Points: 1

# Comments: 0

Article URL: https://www.businessinsider.com/anthropic-changing-safety-policy-2026-2

Comments URL: https://news.ycombinator.com/item?id=47160324

Points: 1

# Comments: 0

Article URL: https://elevenfreedoms.org/

Comments URL: https://news.ycombinator.com/item?id=47160297

Points: 1

# Comments: 0

Article URL: https://www.typequicker.com/average-typing-speed

Comments URL: https://news.ycombinator.com/item?id=47160283

Points: 1

# Comments: 0

Article URL: https://wtfhappened2025.com/

Comments URL: https://news.ycombinator.com/item?id=47160269

Points: 3

# Comments: 0

Article URL: https://medium.com/@brandon_89699/4df2f34cba14

Comments URL: https://news.ycombinator.com/item?id=47160263

Points: 1

# Comments: 0

Article URL: https://github.com/RightNow-AI/openfang

Comments URL: https://news.ycombinator.com/item?id=47160246

Points: 3

# Comments: 1

Article URL: https://stephaniespanjian.com/blog/rag-cost-reduction-replaced-opensearch-s3-in-memory-search

Comments URL: https://news.ycombinator.com/item?id=47160232

Points: 2

# Comments: 1

Article URL: https://www.eff.org/deeplinks/2026/02/tech-companies-shouldnt-be-bullied-doing-surveillance

Comments URL: https://news.ycombinator.com/item?id=47160226

Points: 5

# Comments: 0

Article URL: https://www.mdpi.com/2304-8158/15/4/712

Comments URL: https://news.ycombinator.com/item?id=47160224

Points: 2

# Comments: 0

Article URL: https://github.com/wilwaldon/Claude-Code-Video-Toolkit

Comments URL: https://news.ycombinator.com/item?id=47160222

Points: 1

# Comments: 0

A Unix-inspired shell and RAM filesystem for the Commodore 64 (6502 assembly)

Comments URL: https://news.ycombinator.com/item?id=47160220

Points: 1

# Comments: 0

Article URL: https://martinalderson.com/posts/which-web-frameworks-are-most-token-efficient-for-ai-agents/

Comments URL: https://news.ycombinator.com/item?id=47160216

Points: 1

# Comments: 0

I spent like 2 months building a tool to solve a problem we had: How do you enforce architectural decisions automatically?

Problem: We're a team of ~20 engineers. Started with clean architecture. Now it's... let's just say "creative layering"

Real issues: - 40% of PRs were rejected just for architecture violations - Code review became the bottleneck (architectural review ≠ logic review) - Junior devs didn't understand the implicit rules - No way to catch violations automatically

Solution: architect-linter

It's like ESLint, but for your entire system design. Define rules in architect.json, architect validates imports across your codebase.

Key features: - Multi-language: TypeScript, JavaScript, Python, PHP (all via Tree-sitter) - Multi-architecture patterns: Hexagonal, Clean, MVC - Fast: Written in Rust, parallel processing - Free & open source (MIT license) - Works in CI/CD, pre-commit hooks, watch mode

Example rule: ```json { "forbidden_imports": [ { "from": "src/components/*", "to": "src/services/*", "reason": "UI layer shouldn't call services directly" } ] }

Comments URL: https://news.ycombinator.com/item?id=47160201

Points: 1

# Comments: 0

Article URL: https://robertreich.substack.com/p/pete-hegseth-and-the-ai-doomsday

Comments URL: https://news.ycombinator.com/item?id=47160193

Points: 5

# Comments: 1

Article URL: https://www.nytimes.com/2026/02/22/us/politics/cyber-agency-dhs-security-setbacks.html

Comments URL: https://news.ycombinator.com/item?id=47159912

Points: 1

# Comments: 0

Hi HN, I built TuneJourney.com, a web app that lets you explore ~70k radio stations worldwide via a 3D globe. While sites like Radio Garden are great for discovery, I wanted to build something that functioned more like a modern media player with automation—specifically, the ability to automatically skip a station when it switches from music to talking or ads.

The Unique Part: In-Browser ML The core feature is an "AI Skip Talk." Instead of processing audio on a server, it uses the Web Audio API to capture the stream locally and runs a lightweight classification model directly in your browser. It estimates the "music vs. speech" probability in near real-time.

If you have the filter enabled, it will automatically trigger a "next" command to hop to another station in your playlist or a random location on the globe the moment a DJ, news segment, or ad starts.

Privacy: No audio data ever leaves your machine; inference is entirely local.

The Tech Stack/Features: - WebGL + Point Clustering: To render 70,000 stations across 11,000 locations smoothly. - In-browser Inference: Uses a small model plus stream-level heuristics to handle edge cases. - Media Key Integration: Full support for physical keyboard and system-level Next/Prev buttons. - Persistence: Sign-in is available to sync playlists and favorites across devices, but the core explorer works without an account. - Online activity: You can see in real time other people on the site, where they are on the globe, what they listened to, and what stations they liked. - Simple games to kill the time as you are listening to stations, like Solitaire, Minesweeper, etc.

The Trade-offs & Challenges: Running a WebGL globe and real-time audio analysis simultaneously is heavy on the CPU. To handle this, I’ve implemented: - Automatic performance detection that downgrades visuals on lower-end machines. - A manual toggle to kill the audio processing if you just want to use the globe as a standard player. - A talk sensitivity slider for the ML model, so you can tweak it yourself.

What I’d love feedback on: - Performance: How does your CPU/fan react? Is it manageable? I’m looking for ways to further optimize the client-side ML or perhaps it is okay to bring even heavier models with more accuracy. - Classification Accuracy: How useful is it? Does it detect talking in most of the cases, or is it sometimes useless? On MacBook it works ok-ish. Ads with music are difficult but when music changes to pure talk, the site does its work and hops to another one.

Let me know what you think! I am interested if this project is worth further investment, building a mobile app, etc.

Comments URL: https://news.ycombinator.com/item?id=47159891

Points: 1

# Comments: 0