As AI moves beyond chatbots and toward systems that can take actions, the Linux Foundation is launching a new group dedicated to keeping AI agents from splintering into a mess of incompatible, locked-down products. The group, dubbed the Agentic AI Foundation, will act as a neutral home for open-source projects related to AI agents.
Anchoring the AAIF at launch are donations from Anthropic, Block, and OpenAI. Anthropic is donating its Model Context Protocol, a standard way to connect models and agents to tools and data. Block is contributing Goose, its open-source agent framework. OpenAI is bringing AGENTS.md to the table, its simple instruction file developers can add to a repository to tell AI coding tools how to behave. You can think of these tools as the basic plumbing of the agent era.
Other members in the AAIF include AWS, Bloomberg, Cloudflare, and Google, signaling an industry-level push for shared guardrails so that AI agents can be trustworthy at scale.
In OpenAI engineer Nick Cooper’s view, protocols are essentially a shared language that lets different agents and systems work together without every developer reinventing integrations from scratch. He stated that multiple protocols are needed to negotiate, communicate, and work together to deliver value, and that openness is why it will never be just one provider or one company.
Jim Zemlin, executive director of the Linux Foundation, put it more bluntly. The goal is to avoid a future of closed, proprietary stacks where tool connections and agent behavior are locked behind a handful of platforms. By bringing these projects together under the AAIF, the foundation can coordinate interoperability, safety patterns, and best practices specifically for AI agents.
Block, the fintech company behind Square and Cash App, is making an openness play with Goose. AI Tech Lead Brad Axen frames it as proof that open alternatives can match proprietary agents at scale, with thousands of engineers using it weekly. Open-sourcing Goose serves a dual purpose for Block. It invites outside help to improve the framework, and those improvements flow back to the company. Donating it to the Linux Foundation gives Block access to community stress-tests and positions Goose as a working example of the AAIF’s vision.
Anthropic is making a similar move by handing its Model Context Protocol to the Linux Foundation. The goal is to make MCP the neutral infrastructure connecting AI models to tools and data without endless one-off adapters. According to MCP co-creator David Soria Parra, the main goal is to have enough adoption that it becomes the de facto standard, creating an open integration center where developers build something once and use it across any client. Donating MCP signals the protocol won’t be controlled by a single vendor.
That governance point is central to why the Linux Foundation created this new umbrella. The organization already hosts major AI projects, but the AAIF is specifically aimed at agent standards and orchestration. Its structure is funded through membership dues, but the Linux Foundation argues funding doesn’t equal control, as project roadmaps are set by technical steering committees.
The big question is whether AAIF becomes real infrastructure or just another industry alliance. Zemlin says an early indicator of success would be the adoption and implementation of its shared standards by vendor agents around the world. For OpenAI’s Cooper, success would mean the protocols evolve continually and accept further input, rather than becoming stagnant.
There is also a subtle consequence: even with open governance, one company’s implementation could become the default simply because it gains the most usage. Zemlin points to open-source history, like Kubernetes winning the container race, as evidence that dominance can emerge from merit and not vendor control.
For developers and enterprises, the short-term appeal is clear: less time building custom connectors, more predictable agent behavior, and simpler deployment in secure environments. The larger vision is more ambitious. If tools like MCP, AGENTS.md, and Goose become standard infrastructure, the agent landscape could shift from closed platforms to an open, mix-and-match world reminiscent of the interoperable systems that built the modern web.