Agent Skills: Anthropic Hands the Standard to the Agentic AI Foundation
A million skills published, nearly 280 million installed. From now on the standard will be maintained by individuals from Anthropic, Google and OpenAI.
Claude Desktop 3.6556.1: SSH Hosts From Your Own Command, Key Presses in the iOS Simulator
A folder's permission mode is now stored separately for this computer, each SSH host and each WSL distribution. Folders set to 'Bypass permissions' start back in their default mode.
Formal Verification with Claude: 180 Million Lines of Rocq Across 123 OSS-Fuzz Libraries
One proof ran into a counter underflow that only fires at roughly 20 megabytes of input with a rare layout. Fuzzers usually cap their inputs at one megabyte.
Claude Models on Google Cloud (Anthropic Managed): The Full Claude Platform Through Google Cloud IAM
Fast Mode for Opus 5.5 is generally available there at $8 per million input and $40 per million output tokens. New features and betas land the same day they do on Claude Platform.
METR and Redwood Research Find Flaws in OpenAI's Internal Monitoring System
One employee from each organisation worked on site for three weeks, chaperoned, on OpenAI-issued devices that never left the premises.
The Signal Chats With Ryan Greenblatt: Why OpenAI Fired Three Safety Researchers
OpenAI's security team called the three into meetings in September and asked them to unlock their personal phones.
Walmart Agent Gateway: An MCP Server for Purchases by AI Agents
Agent Profiles arrive next month: one profile per agent inside your account, with its purchases listed separately and its own spending cap.
Attack Surface Reduction: Claude Finds 493 Issues Across 25 Open Source Projects
One pattern showed up in four unrelated projects: a deny rule written with a hostname never fires when reverse DNS is off. None of them documented that.