October 22, 2026 — The Agentic AI Foundation (AAIF) announced today that four major AI companies are handing it key agent projects: Anthropic is contributing Agent Skills, OpenAI is contributing Open Responses, Hugging Face is contributing OpenEnv, and Alibaba is contributing OpenSandbox. The additions bring the foundation to 10 hosted projects in its first year — and they put the plumbing of the agent era under neutral, shared governance instead of any single vendor’s control.
What happened
The announcement was made at AGNTCon + MCPCon North America, taking place October 22–23 at the San Jose McEnery Convention Center. AAIF launched in December 2025 with three projects. The four new ones join an existing roster that already includes the Agent2Agent (A2A) protocol, Agent Router, agentgateway, AGENTS.md, goose, and the Model Context Protocol (MCP) — the widely used standard for connecting AI models to tools and data, which Anthropic donated to the foundation less than a year ago.
“A year ago, AAIF launched with three projects and a bet that agentic AI needed a neutral home to thrive. Today we host 10,” said Manik Surtani, the foundation’s CTO and co-founder.
The four new projects, one by one
| Project | Contributed by | What it does |
|---|---|---|
| Agent Skills | Anthropic | An open format for giving agents new capabilities: instructions, scripts and resources packaged so agents can load specialized know-how on demand, portable across products and platforms. |
| Open Responses | OpenAI | An open specification for multi-provider LLM interfaces, based on the OpenAI Responses API — a shared schema and consistent streaming events so developers describe requests once and run them across providers. |
| OpenEnv | Hugging Face | An open framework and community hub for building reinforcement-learning environments that are interoperable and plug-and-play, exposed through a containerized service using MCP and the Gym training protocol. |
| OpenSandbox | Alibaba | A general-purpose sandbox platform for running AI applications safely, with multi-language SDKs and Docker and Kubernetes runtimes for coding agents, GUI agents, evaluation and RL training. |
Together, the four cover the full agent stack the foundation describes: how agents gain capabilities, how applications talk to models across providers, how agents are trained and evaluated, and where agents run safely.
Why it matters
Standards fights are usually boring until they decide who pays rent to whom. If the formats for agent skills, model interfaces, training environments and sandboxes live in a neutral foundation, a skill or an integration built once can travel across Anthropic, OpenAI, Hugging Face and Alibaba ecosystems instead of being rebuilt — and locked in — per vendor. Anthropic’s David Soria Parra, a co-creator of MCP, made the same argument for the donation: builders who use skills should be able to shape what comes next. For businesses betting on agents, this is the tested-intelligence takeaway: the agent stack is consolidating around open, shared infrastructure faster than most expected, and interoperability is becoming a feature vendors compete to support rather than a favor they grant.
Frequently asked questions
What is the Agentic AI Foundation?
AAIF is a neutral home for open standards, protocols and infrastructure for agentic AI. Its founding projects include MCP, goose and AGENTS.md, and it now hosts 10 projects under open, vendor-neutral governance.
What did Anthropic, OpenAI, Hugging Face and Alibaba contribute?
Anthropic contributed Agent Skills, OpenAI contributed Open Responses, Hugging Face contributed OpenEnv, and Alibaba contributed OpenSandbox, all announced on October 22, 2026.
Does this change anything for developers today?
The projects themselves already exist; what changes is their governance. They will now evolve under the foundation’s shared process, which is meant to keep them interoperable and free of single-company control.
Sources: Agentic AI Foundation announcement via PR Newswire (October 22, 2026); Unite.AI.