Agentic AI Foundation adds Agent Skills, Open Responses, OpenEnv and OpenSandbox
The foundation's project count rises from three to ten; Anthropic says Agent Skills has been integrated by more than 70 platforms, with 1 million skills published and nearly 280 million installs.
- Open weights & ecosystem
- Minor
The Agentic AI Foundation (AAIF), a Linux Foundation body, announced four new projects on 22 October at its AGNTCon and MCPCon North America event in San Jose. Anthropic contributed Agent Skills, OpenAI Open Responses, Hugging Face OpenEnv and Alibaba OpenSandbox. The foundation now hosts ten projects, up from the three it launched with in December 2025.
The release describes each project. Agent Skills is an open format for packaging instructions, scripts and resources that agents load on demand, so that a skill can be reused across products. Open Responses is a specification for multi-provider model interfaces based on OpenAI’s Responses API. OpenEnv is a framework and community hub for reinforcement-learning environments, run as containerised services. OpenSandbox is a sandbox platform for running agents, with multi-language SDKs and Docker and Kubernetes runtimes. The other six projects are the Model Context Protocol (MCP), goose, AGENTS.md, the Agent2Agent protocol, Agent Router and agentgateway.
Anthropic’s own post says the donation includes the Agent Skills name, logo, GitHub organisation and agentskills.io domain. It reports that more than 70 agent platforms, including ChatGPT, Cursor, GitHub Copilot and VS Code, have integrated the standard since it was published in December 2025, with 1 million skills published and nearly 280 million installs. Those figures are Anthropic’s. The company said maintainers, drawn from Anthropic, Google, OpenAI and others, serve “as individuals”, and that the standard can be used or contributed to without joining the foundation. It listed a roadmap of versioned releases, a formal proposal process and security guidance for authors and installers of skills.
Manik Surtani, the foundation’s chief technology officer and co-founder, said in the release that the aim is “one place to build shared standards, protocols, and infrastructure” with “no single company controlling the outcome”.
The Register’s headline framed the move more sceptically, as vendor-affiliated technology being adopted by a foundation those vendors fund. All four projects originate with one of the contributing companies, and the foundation is backed by them and other members. The donations follow Anthropic’s earlier contribution of MCP and move the skills format, which Anthropic introduced in October 2025, from single-company stewardship to the same neutral host.
In the commentary
What people were saying around this time — external links, from the record's commentary rail.
- 22 October 2026 · Buck Shlegeris · Redwood ResearchRed-Teaming OpenAI's Internal Agent Monitoring Systems
- 20 October 2026 · Zvi Mowshowitz · Don't Worry About the VaseAnthropic Does Not Understand Why Its Draft Bill Was Unacceptable and Troubling
- 19 October 2026 · Shakeel Hashim · TransformerAnthropic’s weak AI bill undermines its ‘safety first’ approach