Anthropic announced on October 22 that it is donating Agent Skills, the open standard that gives AI agents specialized knowledge and workflows, to the Agentic AI Foundation under the Linux Foundation. The donation covers the Agent Skills name and logo, the GitHub organization, and the agentskills.io domain. Anyone can keep using or contributing to the standard without joining the foundation, and engineers from Anthropic, Google, OpenAI, and other companies will maintain it as individuals, with David Soria Parra and Mahesh Murag as lead maintainers.
A skill is deliberately simple: a folder containing a SKILL.md file that teaches an agent a specific task or domain, plus any scripts or reference material it needs. Anthropic developed the format while building Claude Code, which handled general knowledge work well but lacked the procedural knowledge real business tasks require. After the October 2025 launch, developers published thousands of skills and wrote loaders for rival tools. Anthropic published the format as an open standard in December 2025, and it now reports more than 70 integrated platforms, including ChatGPT, Cursor, GitHub Copilot, and VS Code, with 1 million skills published and nearly 280 million installed.
Agent Skills did not arrive at the foundation alone. Announced the same day at AGNTCon + MCPCon North America in San Jose, the foundation also took in Open Responses from OpenAI, OpenEnv from Hugging Face, and OpenSandbox from Alibaba. That brings it to 10 hosted projects in its first year, up from three at its December 2025 launch, joining the Model Context Protocol, AGENTS.md, goose, the Agent2Agent protocol, and others. Together the new projects cover how agents gain skills, how applications talk to models across providers, how agents are trained and evaluated, and where they run safely.
For builders, this is the MCP playbook repeated, and that is the point. A format owned by one lab, however open it claims to be, carries a quiet tax: competitors hesitate to build on it and enterprises hesitate to standardize on it. Moving Agent Skills, like the Model Context Protocol before it, into neutral governance removes that hesitation. A skill written once for one agent platform now has a credible path to running everywhere, which makes time spent packaging up your own procedures, checklists, and domain rules a durable asset instead of a platform-specific bet.
Treat that portability with respect, because a skill is not just documentation. It can carry scripts an agent will run, which makes an installed skill a piece of trusted code, not a harmless prompt. Anthropic's proposed roadmap, covering versioned releases, a formal proposal process, and security guidance for people who write and install skills, is therefore the part to watch. Audit skills you inherit the way you audit dependencies, pin the versions you rely on, and write your own as if another team's agent will read them cold. That discipline, not the donation headline, is what will make skills safe infrastructure.
