Feed aggregator

I made this for personal use but thought it may be useful for others as well.

Comments URL: https://news.ycombinator.com/item?id=47137558

Points: 1

# Comments: 0

Article URL: https://scitechdaily.com/scientists-create-chip-that-generates-brand-new-colors-of-light-cracking-a-decades-old-nonlinear-optics-challenge/

Comments URL: https://news.ycombinator.com/item?id=47137540

Points: 1

# Comments: 1

Article URL: https://www.cnbc.com/2026/02/23/ibm-is-the-latest-ai-casualty-shares-are-tanking-on-anthropic-cobol-threat.html

Comments URL: https://news.ycombinator.com/item?id=47137532

Points: 1

# Comments: 0

Article URL: https://www.robert-glaser.de/what-if-iteration-is-all-we-need/

Comments URL: https://news.ycombinator.com/item?id=47137525

Points: 1

# Comments: 0

Hi HN, I'm Micah Blachman, the creator of Crosspassion. After posting about my previous version a few days ago, I rebuilt the entire app from the ground using Codex.

Crosspassion aims to be Reddit, but for finding where your interests overlap. If I was interested in, say, mechanical keyboards and type design, I could check the mechanical keyboard/type design community, and look at the latest discoveries of projects that combine both interests. Plus, you get to connect with people who share your exact crosspassion, or interest combination.

I'm not a developer, so I used Codex for the code. I'm the first to say AI-generated websites and web apps are not a match for the quality of human-created projects, but Crosspassion is still very early-stage.

I'd love feedback on how to improve Crosspassion even further, and if you're a developer and you have any ideas about how to take the project to the next level, feel free to reach out!

I wrote about the idea behind Crosspassion on my newsletter: https://micahblachman.beehiiv.com/p/crosspassion/, and you can contact me at micah.blachman@gmail.com.

-Micah

Comments URL: https://news.ycombinator.com/item?id=47137522

Points: 1

# Comments: 0

I kept running into the same pattern: calling an LLM API thousands of times with the same prompt template, just swapping in different text. Classify this contract clause. Route this support ticket. Categorize this log line. For teams handling contracts, patient records, or internal logs, sending that data to a third-party API isn't always an option. And at scale, you're paying per-token for what's essentially pattern matching. So I built a CLI that trains a small local classifier from labeled examples. Give it 50 input/output pairs, it trains a ~230KB model on your machine, and you run inference locally. No network calls. Everything runs on Node.js - no Python, no GPU, no Docker. Under the hood it uses all-MiniLM-L6-v2 for sentence embeddings (runs locally, downloads once at ~80MB) and trains a small neural network on top of your labels. For topic/domain classification - where categories are about different things - I'm seeing 80-95% accuracy with 50 examples. It struggles with sentiment and tone (44-50%), because "amazing camera" and "terrible camera" produce nearly identical embedding vectors. I documented this openly in the benchmarks. The benchmarks use real text from AG News (127K articles) and 20 Newsgroups (18K posts), with only 50 training samples drawn from each. The test harness and all fixture data are in the repo - clone it and run npx tsx tests/harness/run.ts to reproduce. This isn't trying to replace LLMs. It's specifically for the repetitive classification tasks where the same prompt structure processes different data every time. Open source, Apache 2.0. Still early. Curious whether anyone has tried similar embedding+classifier approaches for their own workflows, or if there's demand for multi-label classification.

Comments URL: https://news.ycombinator.com/item?id=47137506

Points: 1

# Comments: 0

Hey HN,

A few years ago the worst boss I ever had told me “I don’t encourage side projects.”

So obviously I spent a few evenings recently to build Pythia (live beta at https://pythia-rating.com).

It’s a single letter grade (AAA down to C) that combines five indices into one executive-friendly score: •Performance (40 %) – real CrUX field data + Lighthouse lab •Security (20 %) – modern HTTP security headers •Privacy & Tracking (20 %) – cookies, trackers, 3rd-party domains, consent signals •Sustainability (15 %) – page weight, green hosting, CO₂ estimates, image bloat •Infrastructure (5 %) – CDN usage + aggressive caching headers

Plus a fun “Amazon lost revenue” calculator (100 ms delay ≈ 1 % revenue impact, adjustable by your ARR and e-comm %).

Some illustrative point-in-time scans (as of ~13:00 GMT today, public data only, indicative):

AI companies •OpenAI: CCC •xAI: B •Anthropic: CCC •Google DeepMind: BB

(Keeping Elon happy.)

Universities • MIT: CCC • Harvard: CCC • Oxford: BBB • Cambridge: BBB (Oxbridge quietly winning on digital health.)

Everything runs on public sources (CrUX, Lighthouse, custom resilient HTTP agent for headers/cookies). No logins, no tracking on the scanner itself. Built purely as a side project because I was tired of 47-metric dashboards that no CEO ever reads. Wanted one number + dollar impact + actual fix suggestions.

Would love your brutal feedback on: •Scoring weights (do they feel right?) •Missing metrics you’d want to see •Edge cases it gets wrong •Who else this could actually be useful for (agencies doing client audits? PE doing diligence? in-house teams benchmarking competitors?)

Happy to answer questions or add anything people ask for. Thanks for looking!

Comments URL: https://news.ycombinator.com/item?id=47137502

Points: 1

# Comments: 0

Article URL: https://ramansharma.substack.com/p/kafka-can-be-so-much-more

Comments URL: https://news.ycombinator.com/item?id=47137500

Points: 1

# Comments: 0

Article URL: https://thescienceplayground.com/

Comments URL: https://news.ycombinator.com/item?id=47137499

Points: 1

# Comments: 0

Article URL: https://waymo.com/blog/2026/02/dallas-houston-san-antonio-orlando

Comments URL: https://news.ycombinator.com/item?id=47137493

Points: 1

# Comments: 0

Hi HN, I originally started this project simply to safely run autonomous coding agents (like Claude in auto-mode) in isolated environments. But as I built it, I realized the potential was much bigger than just sandboxing: it's about making a coding agent callable like a standard API, and ultimately, defining your entire development workflow as infrastructure.

Axon is a Kubernetes-native framework that abstracts the coding agent itself. By standardizing the container interface, you can seamlessly plug in Claude, Codex, Gemini, OpenCode, or your own custom images without changing your orchestration.

How it works under the hood:

The architecture relies on standard K8s CRDs (Tasks, Workspaces, TaskSpawners, and AgentConfigs).

A Task is the fundamental unit of work for a coding agent. When a Task is created, Axon spins up an ephemeral Pod with a freshly cloned git workspace to execute it.

You use an AgentConfig to declaratively define the specific rules, custom skills, and MCP (Model Context Protocol) servers the agent is allowed to access during that Task.

A TaskSpawner reacts to GitHub events (like a new issue or PR) and automatically generates these Tasks on the fly.

Deterministic outputs (branch names, PR URLs, token usage) are captured directly back into the K8s resource status.

To test this, I built a fully autonomous self-development pipeline. It actively triages issues, suggests new features, finds bugs, and writes the code to fix them. It uses standard GitHub issues and PR comments as the feedback loop to guide the agents. You can see how that is wired together here: https://github.com/axon-core/axon/tree/main/self-development

The main repo is here (with demo video): https://github.com/axon-core/axon

I’d love to get your real-world feedback—especially on the CRD design, the container abstraction model, and what new features you would need to actually integrate this into your workflows.

Comments URL: https://news.ycombinator.com/item?id=47137491

Points: 1

# Comments: 0

Article URL: https://stripe.com/en-at/annual-updates/2025

Comments URL: https://news.ycombinator.com/item?id=47137490

Points: 1

# Comments: 0

Article URL: https://github.com/Miscend/codex-rebuilder

Comments URL: https://news.ycombinator.com/item?id=47137484

Points: 2

# Comments: 0

Article URL: https://docs.julialang.org/en/v1/manual/performance-tips/

Comments URL: https://news.ycombinator.com/item?id=47137479

Points: 1

# Comments: 0

Claude Code recently added native worktree support (--worktree), which helps a lot with parallel agent isolation.

What's still unsolved is conflicts across active worktrees, which often surfaces late.

I built Clash (OSS, Rust) to detect potential conflicts earlier using read-only merge simulation across worktrees during edits.

- clash check for pre-edit checks - clash status for a cross-worktree conflict matrix - clash watch for live monitoring - Claude plugin hook support for Write|Edit|MultiEdit Quick setup (CLI + Claude plugin):

curl -fsSL https://clash.sh/install.sh | sh && \ (claude plugin marketplace add clash-sh/clash || true) && \ (claude plugin install clash@clash-sh || claude plugin update clash@clash-sh) Would love feedback or your thoughts on this.

Comments URL: https://news.ycombinator.com/item?id=47137470

Points: 1

# Comments: 0

Article URL: https://www.nytimes.com/2026/02/24/business/meta-amd-chips-ai.html

Comments URL: https://news.ycombinator.com/item?id=47137467

Points: 1

# Comments: 0

Timothy Youngblood was CISO at Dell, CISO at Kimberley-Clark, VP & CISO at McDonald’s, and SVP, CSO & Product Security Officer at T-Mobile.

The post CISO Conversations: Timothy Youngblood; 4x Fortune 500 CISO/CSO appeared first on SecurityWeek.

From AI hardware to wearable phones, these products promised a lot. So what happened to them?

The malicious code propagates like a worm, poisons AI assistants, exfiltrates secrets, and contains a destructive dead switch.

The post New ‘Sandworm_Mode’ Supply Chain Attack Hits NPM appeared first on SecurityWeek.