frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

The Anthropic Hive Mind

https://steve-yegge.medium.com/the-anthropic-hive-mind-d01f768f3d7b
1•gmays•2m ago•0 comments

The Sling: Humanity's Forgotten Power

https://www.slinging.org/
1•jsattler•7m ago•1 comments

Noam Chomsky, Jeffrey Epstein and the Politics of Betrayal

https://chrishedges.substack.com/p/noam-chomsky-jeffrey-epstein-and
2•chmaynard•8m ago•0 comments

Why securing AI model weights isn't enough

https://www.the-substrate.net/p/why-securing-ai-model-weights-isnt
1•erwald•8m ago•0 comments

Cursor Composer 1.5

https://cursor.com/blog/composer-1-5
4•leerob•10m ago•1 comments

Stop Telling Users Their DNS Is Wrong

https://jacob.gold/posts/stop-telling-users-their-dns-is-wrong/
2•jacobgold•10m ago•0 comments

Show HN: Voice Legacy: AI that interviews your parents before it's too late

https://www.voicelegacy.app
1•blacksausage•10m ago•1 comments

Have the patents for H.264 MPEG-4 AVC expired yet?

https://meta.wikimedia.org/wiki/Have_the_patents_for_H.264_MPEG-4_AVC_expired_yet%3F
1•Velocifyer•12m ago•0 comments

The Great Displacement: AI and the Next Fifty Years

https://drive.google.com/file/d/1nm0wOCJJvGS0WI737XHJ7VVDkjQrbgXf/view
1•sideway•13m ago•0 comments

GitButler CLI Is Good

https://matduggan.com/gitbutler-cli-is-really-good/
1•weaksauce•13m ago•0 comments

Modern Keystroke Visualizer for Linux

https://github.com/linuxmobile/keystroke
1•lwhsiao•14m ago•0 comments

AskHN: Is Auth0 Down Again?

2•jansan•14m ago•0 comments

Strengthening Windows trust and security through User Transparency and Consent

https://blogs.windows.com/windowsexperience/2026/02/09/strengthening-windows-trust-and-security-t...
1•pentagrama•16m ago•0 comments

Paragraphic – Parametric graphic design app made in Godot

https://paragraphic.design/
1•count_zero•18m ago•0 comments

Ask HN: I experienced an Attack on Telegram and simcards gone!!!

2•khoobid_shoma•20m ago•0 comments

Ask HN: Any good open source projects written by AI agents?

2•grillorafael•20m ago•0 comments

European Processor Initiative

https://www.european-processor-initiative.eu/project/epi/
2•Gravityloss•21m ago•0 comments

Show HN: Linkpreview.io – Debug and preview social share cards

https://linkpreview.io
1•ravikmd•21m ago•0 comments

The power of anime: using anime for education and outreach in STEM

https://www.frontiersin.org/journals/education/articles/10.3389/feduc.2025.1707055/full
2•PaulHoule•23m ago•0 comments

German patent classified as state secret

https://zenodo.org/records/18551454
2•nAOpx•23m ago•1 comments

Show HN: MumbleFlow – $5 local voice-to-text (whisper.cpp, Rust, no cloud)

https://mumble.helix-co.com
1•mumbleflow•23m ago•0 comments

Hims cancels plans to sell compounded GLP-1 pill after FDA backlash

https://www.biopharmadive.com/news/hims-stop-launch-compounded-wegovy-fda-novo/811662/
2•randycupertino•23m ago•2 comments

Regime-Declared Mathematics as Survivor Sets

https://zboralski.github.io/br/maths/index.html
1•kugutsumen•24m ago•0 comments

National data, local stories: ICE detention in 2026

https://exclav.es/2026/02/08/exploring-ice-detention-facilities/
2•ai_critic•24m ago•1 comments

Is AI the Paperclip?

https://www.newcartographies.com/p/is-ai-the-paperclip
3•headalgorithm•28m ago•0 comments

AGI/Singularity: 9,300 Predictions Analyzed

https://research.aimultiple.com/artificial-general-intelligence-singularity-timing/
1•hakkikonu•29m ago•0 comments

America Has a Tungsten Problem

https://www.noleary.com/blog/posts/1
19•noleary•32m ago•3 comments

Show HN: A last-minute romantic gift app with private links

https://fiveminutelove.com/
1•shivamjjha•34m ago•0 comments

Show HN: I killed my Calendly link after people booking randomly

3•Mrakermo•35m ago•1 comments

Show HN: Claude Cowork for Startup Market Analysis

https://brainwave.vc/prompt
1•louison11•40m ago•0 comments
Open in hackernews

Show HN: SDF Protocol – Pre-compiled semantic JSON for AI agent web consumption

https://sdfprotocol.org/
1•spranab•1h ago

Comments

spranab•1h ago
Hi HN, I built SDF (Structured Data Format), an open protocol that sits between web content and AI agents.

The problem: Every agent that consumes a web page independently fetches HTML, strips boilerplate, extracts entities, and classifies content. A typical page is ~89KB of HTML (~73K tokens). When 100 agents consume the same URL, this extraction happens 100 times with inconsistent results.

What SDF does: Convert once into a schema-validated JSON document (~750 tokens) containing entities, claims, relationships, summaries, and type-specific structured data. Agents consume the pre-extracted representation directly.

Results from production deployment (2,335 documents, 10 content types):

99.2% token reduction from HTML 90% extraction accuracy with fine-tuned 1.5B + 3B model cascade 4.1x faster than monolithic 14B baseline Downstream experiment: general-purpose 7B model scores 0.739 accuracy from SDF vs 0.352 from raw markdown (p < 0.05) The pipeline runs locally on consumer hardware (dual RTX 3090 Ti). Fine-tuned models are open on HuggingFace (sdfprotocol/sdf-classify, sdfprotocol/sdf-extract). Protocol spec and JSON schemas are on GitHub.

Protocol spec + schemas: https://github.com/sdfprotocol/sdf Whitepaper: https://doi.org/10.5281/zenodo.18559223 Models: https://huggingface.co/sdfprotocol Happy to answer questions about the design decisions, the type system, or the evaluation methodology.

ksaj•1h ago
I wonder if people will eventually surf ad-free by sniffing out these files. Easy to parse (maybe even easier than the actual article itself) and no ads or otherwise unrelated distractions.
spranab•1h ago
What do you mean? I just wanted to share something I am working on. Trying to understand what you meant by ads.
ksaj•52m ago
Not your ads. I'm saying that if a site that has ads also has these files, you could get the gist of the article by reading these files instead of going to the ad-laden page itself.
spranab•51m ago
That actually is great, we can add ads detection and extract only the relevant information. Thanks @ksaj
ksaj•46m ago
That's a step further than I was thinking, but I most definitely like the direction.