By Hamza Faisal
Anthropic is giving away one of the most widely adopted building blocks of the AI agent era. The company announced on October 22 that it is donating Agent Skills, the open standard for packaging reusable instructions and workflows for AI agents, to the Agentic AI Foundation (AAIF), an organization under the Linux Foundation.
The transfer is total: the Agent Skills name and logo, the project's GitHub organization, and the agentskills.io domain all move to the foundation. Developers can keep using the standard, or contributing to it, without joining the AAIF, Anthropic said.
The standard that taught agents to follow procedures
A skill is a folder containing a SKILL.md file — a document that teaches an agent how to perform a specific task or work in a specific domain — along with any scripts and reference material the task needs. Instead of re-explaining a team's procedure in every chat, a developer packages it once, and any compatible agent loads it on demand.
The idea grew out of Anthropic's work on Claude Code. The coding tool handled general knowledge work well, the company said, but often lacked the procedural knowledge that real business tasks demand: the checklists, conventions, and step-by-step workflows humans follow without thinking. Skills were Anthropic's answer, launched with Claude Code in October 2025 and published as an open standard that December.
Adoption exploded. More than 70 agent platforms have integrated Agent Skills, including ChatGPT, Cursor, GitHub Copilot, and VS Code. Developers have published roughly one million skills, which Anthropic says have been installed nearly 280 million times. Within weeks of the 2025 launch, developers were already publishing thousands of skills on GitHub and writing loaders for Codex, Cursor, and OpenCode.
A neutral home, with a security to-do list
The AAIF is familiar territory for Anthropic: the company co-founded the foundation with Block and OpenAI in 2025, and it previously donated its other major open standard, the Model Context Protocol (MCP), there too. Going forward, engineers from Anthropic, Google, OpenAI, and other companies will maintain Agent Skills in their individual capacities, with David Soria Parra and Mahesh Murag named as lead maintainers.
The proposed roadmap starts with versioned releases, a formal proposal process, and — notably — security guidance for people who write and install skills. That last item is not theoretical: security firm Snyk previously identified 76 malicious skills among samples on ClawHub, a community skills directory. Because a skill can bundle executable scripts, Anthropic recommends auditing unfamiliar scripts, dependencies, and network connections before installing one.
What the handoff changes
In practice, little changes for users today, and that is the point. The donation shifts Agent Skills from a standard led by one company to one governed by an industry foundation, much as MCP's earlier move did. The two standards now sit side by side: MCP connects agents to external tools and data, while Skills define how agents actually do their jobs.
For the fast-growing agent economy, the message is that the plumbing is going open. With a million skills already in the wild and the biggest coding platforms on board, Anthropic is betting that a vendor-neutral home will carry the standard further than its own stewardship could.
By Hamza Faisal
