Y Combinator CEO Garry Tan just open-sourced gstack — a Claude Code toolkit that transforms a single coding agent into a coordinated team of 8 specialist agents, each optimized for a specific phase of the software development lifecycle. He reportedly merged 100 pull requests in 7 days using it. Product Hunt is calling it “God Mode” for developers. Here’s what it is, why it works, and how to set it up. ...

Winning a math olympiad is impressive. Doing original mathematical research — navigating thousands of papers, formulating novel conjectures, constructing long-horizon proofs — is something else entirely. Until recently, that gap separated AI benchmarks from actual scientific contribution. Google DeepMind’s Aletheia is the first system to credibly cross it. Deployed in December 2025 against a database of 700 open mathematical problems, Aletheia autonomously solved four open Erdős problems — longstanding conjectures that professional mathematicians had left unresolved. It didn’t just compute answers; it generated, verified, and revised proofs in natural language, with results that have contributed to peer-reviewed publications. ...

OpenClaw shipped v2026.3.13 early this morning, and it’s a meaningful release for anyone running browser automation, using OpenClaw on mobile, or hitting dashboard performance walls on heavy agentic runs. The headline feature is the Chrome DevTools Protocol (CDP) attach mode — a native integration that lets OpenClaw connect directly to a signed-in live Chrome session for real-time debugging and automation. That’s a significant capability addition: instead of spinning up an isolated browser profile, you can now attach to the Chrome instance you’re already using, complete with your active sessions, cookies, and extensions. ...

The headline number is uncomfortable: 87%. That’s the share of pull requests containing at least one security vulnerability when AI coding agents — Claude Code, OpenAI Codex, and Google Gemini — were used to build real applications from scratch. That’s the finding from DryRun Security’s inaugural Agentic Coding Security Report, published this week and already making waves through security and developer communities. This isn’t a synthetic benchmark. DryRun tested three leading AI coding agents building two real applications each, generating approximately five pull requests per agent. The result: 143 total vulnerabilities documented across 30 pull requests. Nearly nine out of ten PRs had at least one problem. The two leading failure modes were access control gaps and improper token handling. ...

Alibaba just made agentic AI accessible to anyone with a smartphone. And in doing so, it’s turned what was already a heated competition among China’s tech giants into a full sprint. JVS Claw, Alibaba’s new iOS and Android application, allows users to install and deploy OpenClaw AI agents in minutes — no coding required, no command line, no developer setup. The app is free for the first 14 days. It’s a direct play for the hundreds of millions of Chinese mobile users who are curious about AI agents but have no technical background, and it lands at exactly the moment that agentic AI has become a household conversation in China. ...

The PC industry has officially gained a new category. AMD announced today that it is formally introducing the Agent Computer — a class of always-on desktop hardware built specifically to run AI models and autonomous agents locally, 24 hours a day, seven days a week, without any dependency on cloud infrastructure. It’s a significant moment. Hardware categories don’t get invented every year, and when they do, it’s usually because software demand has outpaced existing form factors. In this case, the software is agentic AI, and the demand is real: developers, researchers, and enterprises increasingly want AI agents running persistently — not as cloud calls with per-token billing, but as local processes that are always available, always private, and always free from inference cost. ...

If you’re building production AI agents in 2026, you’ve almost certainly encountered both MCP (Model Context Protocol) and Agent Skills as architectural options. Both are ways to extend what an AI agent can do — but they operate at fundamentally different levels of abstraction, and choosing between them (or combining them) is one of the most consequential architectural decisions you’ll make early in a project. This guide breaks down how each approach works, when each excels, the compatibility patterns for using both together, and the production deployment tradeoffs that practitioners are discovering in the field. ...

Running Claude Code in a Docker container isn’t just a development curiosity — it’s increasingly the recommended way to work with AI coding agents in a way that’s both powerful and secure. Docker published an official guide this week walking through the full workflow: local model execution with Docker Model Runner, real-world tool connections via MCP servers, and securing agent autonomy inside isolated sandboxes. This guide synthesizes that walkthrough into a practical tutorial for developers who want to get running quickly. ...

China has quietly become the world’s largest OpenClaw market — surpassing US usage figures — and the economic ripple effects are transforming the country’s AI industry into something resembling a gold rush. A new WIRED investigation documents what’s happening on the ground: ordinary people renting cloud servers to run OpenClaw agents, buying AI subscriptions in bulk, and driving demand for the lower-cost Chinese AI models that make the economics of running agents feasible at scale. The primary beneficiaries aren’t the users themselves — it’s the cloud providers, AI subscription platforms, and model vendors cashing in on the frenzy. ...

In the fast-moving world of agentic AI tooling, few stories are more striking than what’s happening at Cursor. The AI-powered coding assistant startup is reportedly in talks with investors for a new funding round that would value the company at approximately $50 billion — nearly double the $29.3 billion valuation it secured just last fall. That’s not a typo. In less than six months, Cursor may have doubled its worth on paper. ...