OpenAI co-founded Agentic AI Foundation and donates AGENTS.md | Keryc
OpenAI, along with Anthropic and Block, announced the creation of the Agentic AI Foundation (AAIF) under the Linux Foundation, backed by Google, Microsoft, AWS, Bloomberg and Cloudflare. The key news: OpenAI donates AGENTS.md, an open and simple format for giving project instructions to agents, and hands it to a neutral foundation to ensure its development and adoption.
What OpenAI announced and who is involved
The AAIF is born to provide neutral governance and interoperable infrastructure as agentic systems move from experiments to tools that do real work. It's not just OpenAI: Anthropic and Block are co-founders, and major players like Google, Microsoft and Amazon support the initiative under the Linux Foundation umbrella.
Why does that matter? Because the Linux Foundation has experience maintaining critical projects like Kubernetes and PyTorch, and can help prevent a standard from falling under the control of a single company.
What is AGENTS.md and why it matters
AGENTS.md is a lightweight Markdown file intended to live alongside README.md. Its purpose is simple: give agents project-specific instructions and context, for example code conventions, build steps, or testing requirements.
Why can a text file change the game? Because it makes agent behavior much more predictable when working across different repositories and toolchains. Since its launch in August 2025, AGENTS.md has been adopted by more than 60,000 projects and agent frameworks, including Amp, Codex, Cursor, Gemini CLI and GitHub Copilot.
OpenAI has also contributed key building blocks to the ecosystem: the Agents SDK, the Apps SDK, the Agentic Commerce Protocol, gpt-oss models and the Codex CLI tool. That last one has helped merge over two million public pull requests on GitHub — a number that shows the practical scale of these components.
A practical example
Imagine your project has an AGENTS.md with: build commands, lint rules, test steps and mock credentials. An assistant that can read that file will run tests in the right order, follow style rules and avoid actions that would break the repository. Reduced errors, faster automated integrations — sounds useful, right?
Why the foundation and neutrality matter
As agents start taking on real responsibility in companies and products, fragmentation will be costly. Without common conventions and neutral governance, we risk incompatible silos that limit portability, security and progress.
The AAIF aims to avoid that: create a space where interoperability standards for agents evolve openly, with contributions from tools, companies and individual developers.
What this changes for developers and companies
For developers: a portable, simple format (AGENTS.md) that makes it easier to build agents that behave well across projects.
For companies: greater predictability and the ability to audit how agents act inside critical workflows.
For the open source community: a process to evolve the format with many voices, not under exclusive control of one company.
There are legitimate questions: how will changes to the standard be managed? what security and privacy mechanisms will the foundation have? The answer lies in the Linux Foundation's governance and the open invitation to contribute.
What you can do now
If you're a developer, maintainer or researcher, follow the evolution of AGENTS.md and consider adding it to your repositories to improve interoperability with assistants and agents. If you work at a company, it's a good time to evaluate how agents could be integrated securely and reproducibly into your processes.
The arrival of the AAIF isn't the end of the road, but it marks an important point: the ecosystem is betting on open standards so agents can scale without losing compatibility or community control.
