Anthropic and OpenAI hand their agent standards to a neutral body
Companies
Agents
Agent Skills and Open Responses join the Agentic AI Foundation, so a skill or a model call can work the same across companies.
What happened
- On October 22, at the AGNTCon conference in San Jose, the Agentic AI Foundation, a neutral industry body set up in December 2025, took in four projects: Agent Skills from Anthropic, Open Responses from OpenAI, OpenEnv from Hugging Face and OpenSandbox from Alibaba. It now hosts ten projects, among them the MCP standard for connecting tools and the A2A standard for agents talking to each other.
- Agent Skills is a format for a folder of instructions, scripts and files that an agent loads only when a task needs it. Open Responses is one shared way to call models from different companies, based on OpenAI’s own interface.
- Both already existed. What changes is governance: no single company controls them any more.
Why it matters for you
A standard that no vendor owns lowers switching costs. A skill written for Claude should work in other agents; code written against one model’s interface should run on another with little change. Less lock-in means more price competition, which is good for users. For your own work: put a recurring task (a data pipeline, a paper’s format, a video workflow) into a skill folder once. It can then travel with you from one tool to the next.
Sources
- Agentic AI Foundation announces four new projects, Agentic AI Foundation, October 22, 2026.
- Agentic AI Foundation adds four new open agent projects, Techstrong.ai, October 23, 2026.
Notes for learning. Nothing here is investment advice. All news