Anthropic transfers Agent Skills governance to the Agentic AI Foundation
Anthropic is donating the Agent Skills name, logo, domain and GitHub organization to the Agentic AI Foundation. Engineers from several companies will maintain the standard as individuals. The transfer changes governance of an existing format for task instructions; using it or contributing remains open without foundation membership.
Artificial Intelligence··Night
The transfer covers the name, domain and repository organization
Anthropic is donating Agent Skills to the Agentic AI Foundation, the Linux Foundation’s home for agent technologies. The October 22 announcement covers the standard’s name and logo, GitHub organization and agentskills.io domain. Users can keep using the format and contributing without joining the foundation. Agent Skills supplies procedural knowledge to software agents: a package supplies instructions for carrying out a particular task rather than supplying another general-purpose model. Anthropic developed the approach while working on Claude Code, its coding agent. The company introduced skills in October 2025 and published the open standard in December that year.[1], [2]
Instructions load when a matching task arrives
A skill folder contains a SKILL.md file with at least a name and description. It can also hold scripts, reference documents and assets such as templates. The agent first reads names and descriptions, loads detailed instructions when a task matches, then accesses further resources as needed. That staged process is intended to limit context consumption when many skills are installed. Anthropic lists ChatGPT, Cursor, GitHub Copilot and VS Code among platforms integrating the format. Its own adoption counts exceed 70 platforms, 1 million published skills and nearly 280 million installations; the independent Japanese coverage does not verify the counting method.[1], [2]
Maintainers will contribute as individuals
Engineers from Anthropic, Google, OpenAI and other companies will maintain the standard in an individual capacity. David Soria Parra and Mahesh Murag are named lead maintainers. The proposed roadmap includes versioned releases, a formal process for specification changes and security guidance for authors and installers. Community feedback is planned through the project’s Discord channel, with transparent decisions. Those items remain proposals; an implementation timetable has not yet been established. Anthropic established the foundation with Block and OpenAI in 2025. Model Context Protocol, the protocol connecting agents to tools and data, had already been donated to the same organization; Agent Skills is a separate standard entering that governance structure.[1], [2]
