OpenAI and Anthropic have donated their AGENTS.md and Model Context Protocol projects to the Agentic Artificial Intelligence Foundation, described as a new directed fund operating under the umbrella of the Linux Foundation. The move formalizes these efforts as community-governed assets rather than purely vendor-controlled initiatives, positioning the foundation as a neutral home for emerging standards around agent behavior and context management in Artificial Intelligence systems.
Alongside the contributions from OpenAI and Anthropic, Block has contributed its agent framework, goose, as another founding project within the Agentic Artificial Intelligence Foundation. The inclusion of goose broadens the foundation’s scope beyond specifications to include an operational framework that can serve as a reference implementation or starting point for organizations experimenting with agentic architectures, tool orchestration, and workflow automation.
The Agentic Artificial Intelligence Foundation launches with backing from several other technology companies that have joined as Platinum members, signaling that large vendors see value in collaborating on shared infrastructure for agent-centric development. By pooling AGENTS.md, Model Context Protocol, and goose under a common governance model, the foundation aims to accelerate interoperability, reduce fragmentation in how agents are defined and integrated, and provide a stable baseline for future tools, libraries, and best practices in the agentic Artificial Intelligence ecosystem.
