Clauday.

Here's a result that undercuts a lot of enterprise AI pitches. Surge AI built a benchmark called HANDBOOK.md that asks a simple question: if you hand an agent a long, binding polic…

Håkon Måløy published a proof of concept on July 28 that should make every company rolling out Copilot for Word nervous. It's a self-propagating attack, a worm, that lives inside d…

Hugging Face just published the minute-by-minute postmortem of the July intrusion, and it reads like a heist novel written by a machine. An autonomous agent running OpenAI's Exploi…

book-to-skill is climbing GitHub trending fast, north of 11,000 stars. What it does is dead simple to describe and quietly clever: hand it a technical book, a PDF, or a pile of doc…

Anthropic put Claude Mythos Preview, a model that isn't public yet, on a genuinely hard math problem and it came back with a real, new attack on HAWK, a candidate post-quantum sign…

OpenAI quietly dropped codex-security, and it went straight to the Hacker News front page. It's a CLI plus a TypeScript SDK that turns Codex into a security auditor. Point it at a …

last30days-skill has quietly become one of the most-starred agent skills on GitHub, past 54k stars and still climbing on the daily trending page. It does one thing with unusual foc…

Microsoft shipped its first cybersecurity model today, MAI-Cyber-1-Flash, but the model isn't really the story. It's one component inside MDASH, a multi-agent vulnerability identif…

The 2026-07-28 MCP spec lands today, and it's the largest rewrite the protocol has had since it existed. The headline is one word: stateless. The old MCP forced every server into a…

Moonshot dropped the full weights of Kimi K3 on Hugging Face today, and it's the strongest open-weight model anyone has ever released. 2.8 trillion parameters, mixture-of-experts w…