Case study
Four agent projects move into the Agentic AI Foundation, including Anthropic's Skills standard
What happened
On October 22, the Agentic AI Foundation announced four additions: Anthropic's Agent Skills, OpenAI's Open Responses, Hugging Face's OpenEnv and Alibaba's OpenSandbox. They concern reusable task instructions, common model interfaces, training environments and isolated places for agents to execute code. The foundation's total reaches ten projects; it was not newly created this week. Foundation announcement ↗.
Anthropic specifies the transferred assets: the Agent Skills name and logo, GitHub organization and agentskills.io domain. People can use and contribute without joining the foundation. Maintainers from several companies act as individuals; a formal proposal process is still on the proposed roadmap. Anthropic's account ↗. Open Responses lets an application reuse a request format across compatible providers or gateways, rather than maintain separate adapters. It does not itself move stored customer data, credentials or permissions.
Why it matters
Transferring names, domains and project stewardship to a foundation creates a possible alternative to unilateral vendor control. Common request formats can ease switching among compatible services, but do not transfer stored data, credentials or permissions. Outside contributors’ influence depends on actual decision rules.
