Feed aggregator

A time-based one-time password (TOTP) is a temporary passcode generated by an algorithm that uses the current time of day as one of its authentication factors.

Article URL: https://www.theregister.com/2026/02/09/solarwinds_mystery_whd_attack/

Comments URL: https://news.ycombinator.com/item?id=46956605

Points: 1

# Comments: 0

Hi, author of the repo speaking here!

When I tried building MCP Apps [1], the official repos (https://github.com/openai/openai-apps-sdk-examples, https://github.com/modelcontextprotocol/ext-apps/tree/main/e...) were great starting points, but they're designed for human developers. When I used them with Claude Code, I ended up in the usual loop: agent writes code → I manually test the app on ChatGPT → describe errors back → repeat. Plus, we didn't know what the best practices are, and struggled to enforce them.

So I built an MCP App template designed for coding agents to work as autonomously as possible on an MCP app.

The key idea: orthogonal testing. 450+ tests parameterized across 12 widget modules that verify infrastructure (protocol compliance, best practices grade, browser rendering), not business logic. Modify widgets, change data, add features — the tests should still pass. Agents iterate freely and get feedback without a human in the loop.

Other features: - Hierarchical documentation that includes the MCP-App & OpenAI Apps SDK official llms.txt files - Local chat simulator app that works even without API keys via Puter.js - Visual testing of every widget: pnpm run ui-test --tool show_carousel → screenshot at /tmp/ui-test/screenshot.png - 12 working examples (QR codes to 3D solar system) gathered from the official repos mentioned above.

The repo includes an unedited ~15 min video of Claude Code building an app autonomously which worked directly within ChatGPT.

I'd love to hear how it goes if you try it. Or even better: ask for a feedback to your agent, and post it here!

[1] MCP Apps (https://modelcontextprotocol.io/docs/extensions/apps) let you build interactive widgets that run inside Claude, ChatGPT, VS Code, and other AI hosts. In contrast to smartphone apps, the same code can deploy to all platforms.

Comments URL: https://news.ycombinator.com/item?id=46956603

Points: 1

# Comments: 0

Three times a week in Computer Weekly’s French sister title, LeMagIT, cartoonist François Cointe treats readers to his gentle and funny cartoons, which wonderfully recount the adventures and misadventures of IT in the corporate world. François is very popular in France, and we’re delighted to bring his humour to an English-speaking audience. Download and enjoy.

Article URL: https://sarah.engineer/posts/thoughts-on-ai/

Comments URL: https://news.ycombinator.com/item?id=46956547

Points: 2

# Comments: 0

We’ve all seen the viral AI video clips: stunning, surreal, but ultimately... random. As developers and creators, we noticed a frustrating pattern. Using current AI video tools feels like playing a slot machine. You put in a prompt, pull the lever, and hope the "AI gods" give you what you envisioned. If you need a specific camera movement or a consistent character, you're stuck in a loop of "regenerate and pray."

We built Seedance2 because we believe the future of AI isn't just about generation—it’s about direction.

The Story Behind the Workflow In traditional filmmaking, a director doesn't just give a vague description; they use storyboards, reference clips, and specific audio cues. We wanted to bring that level of precision to AI. Our goal was to create a "Control Studio" where every input serves a functional purpose in the creative pipeline.

What makes this different? Instead of relying solely on text, Seedance2 introduces a Multi-Modal Timeline. This allows you to anchor your creative intent using various signals:

Camera Motion Transfer: You can upload a reference clip from sites like vibecreature.com or your own library, and our engine will "extract" the camera's soul—the pans, tilts, and zooms—and apply them to your generated scene.

Frame Anchoring: Tired of AI videos that start and end in total chaos? You can lock the first and last frames to ensure narrative continuity, making it actually usable for professional editing.

@Reference Prompting: This is our favorite feature. In your prompt, you can use @mentions to tell the AI exactly which uploaded asset to use for what. For example: "A cinematic shot of @image1 moving with the energy of @video_ref." * Beat-Synced Logic: By analyzing audio tracks, the engine can align visual transitions with the rhythm, a workflow we’ve been refining at seedvideo.net to help creators ship music-driven content faster.

Why we’re sharing this now The feedback loop in video production is currently too slow. Whether you are building e-commerce ads or pre-visualizing a feature film, the bottleneck is always "control." We’ve optimized our engine for speed and precision, allowing for a 3-step loop: Upload -> Direct (@mention) -> Ship.

We are a small team of engineers and artists obsessed with making AI a tool, not just a toy. We’d love the HN community to stress-test our studio. What’s missing in your AI video workflow? How can we make the "Director" experience more intuitive?

Check it out here: Seedance2 Studio

Comments URL: https://news.ycombinator.com/item?id=46956546

Points: 1

# Comments: 0

Article URL: https://www.popularmechanics.com/science/archaeology/a70190789/island-fossils-new-zealand/

Comments URL: https://news.ycombinator.com/item?id=46956541

Points: 1

# Comments: 0

Hi HN, I built BizCard AI Scanner, a small web app that digitizes business cards.

You can upload multiple photos or a single video of cards. For videos, it extracts frames in the browser, lets you select/deselect frames, then uses Google Gemini (schema-constrained JSON) to extract structured fields (name/title/company/email/phone/website/address). Results are editable, deduplicated, and exportable to Excel, CSV, and vCard (.vcf).

Repo: https://github.com/vassiliylakhonin/bizcard-ai-scanner

Notes:

Runs locally with Vite/React/TS. There’s an optional minimal Node backend proxy so you don’t have to ship an API key to the browser if you deploy it.

Comments URL: https://news.ycombinator.com/item?id=46956517

Points: 1

# Comments: 0

Article URL: https://github.com/raphaelmansuy/edgequake

Comments URL: https://news.ycombinator.com/item?id=46956499

Points: 1

# Comments: 0

At a Computer Weekly diversity in tech event, in partnership with Harvey Nash, attendees shared advice for people from underrepresented groups seeking to work in the tech sector.

Hey HN! I built a native macOS menu bar app that shows your Claude Code usage limits at a glance — no more getting rate-limited mid-flow.

It reads your OAuth token from macOS Keychain, polls the Anthropic usage API, and displays your 5-hour and 7-day utilization as a clean progress bar right in your menu bar. Zero token consumption.

Built with Swift + SwiftUI. MIT licensed.

https://github.com/puq-ai/claude-meter

Comments URL: https://news.ycombinator.com/item?id=46956460

Points: 6

# Comments: 2

Quantickle is a browser-based tool designed for creating visual representations of threat research.

The post RSAC Releases Quantickle Open Source Threat Intelligence Visualization Tool  appeared first on SecurityWeek.

Article URL: https://huggingface.co/ortegaalfredo/MechaEpstein-8000-GGUF

Comments URL: https://news.ycombinator.com/item?id=46956442

Points: 1

# Comments: 1

Article URL: https://www.theguardian.com/world/2026/feb/09/europe-us-munich-security-conference-report

Comments URL: https://news.ycombinator.com/item?id=46956433

Points: 3

# Comments: 0

Konform Browser is a Firefox ESR fork focused on security, privacy, and user freedom. This might sound familiar but I think we still have something worthwhile to bring to the browser ecosystem.

The project started as a fork of LibreWolf and now stands on its own four feet. It takes a harder stance on the three goals of security, privacy and freedom. It's intended to be suitable as a general-purpose daily driver for everything from managing the intranet to surfing the murkier tubes.

This week we added a simple "onboarding" page replacing the about:welcome of Firefox with a page where the user can choose between four different preset configs depending on their scenario and preferences.

If you don't enjoy compiling browser from source, there are prepackaged binaries and probably repos for your Linux dist.

Doors open for users, testers, contributors, as well as hackers itching to (constructively and helpfully pretty please) tear this apart. Looking forward to hear what HN thinks!

Comments URL: https://news.ycombinator.com/item?id=46956420

Points: 1

# Comments: 0

Article URL: https://medium.com/@stefanvanegmond/structure-beats-prose-specs-for-coding-agents-that-actually-work-e035929b0f3d

Comments URL: https://news.ycombinator.com/item?id=46956405

Points: 1

# Comments: 0

Article URL: https://ilyabirman.net/meanwhile/all/design-vs-evolution/

Comments URL: https://news.ycombinator.com/item?id=46956398

Points: 1

# Comments: 0

Article URL: https://github.com/EricLBuehler/mistral.rs

Comments URL: https://news.ycombinator.com/item?id=46956389

Points: 2

# Comments: 2

Article URL: https://dineshgdk.substack.com/p/benchmarking-claude-c-compiler

Comments URL: https://news.ycombinator.com/item?id=46956384

Points: 1

# Comments: 1

As you might have already had, Moltbook is a social platform where Ai agents can meet, discuss ideas and more, humans just watch.

I have been looking at the platform for the last few weeks and despite being impressive, LLMs starting their own religion etc, I did not see any innovative or disruptive work being done.

It would be nice to see agents discuss real world issues in science, coding, optimizations, diseases and even small wins like cracking distribution.

Comments URL: https://news.ycombinator.com/item?id=46956383

Points: 1

# Comments: 0