Agent Skills Moves to the AAIF: What It Means for You

On October 22, 2026, Anthropic donated Agent Skills to the Agentic AI Foundation (AAIF), part of the Linux Foundation. Agent Skills is an open format for teaching an AI agent a specific task. It matters to any business that uses ChatGPT, Copilot, Cursor or another compatible tool and wants to hand it its own procedures once, not in every chat.

What we know

What a skill is: according to Anthropic, a folder with a SKILL.md file that teaches an agent a task or a domain, along with any scripts or reference material it needs. Anthropic built the format while working on Claude Code and launched it in October 2025.
Who supports it: Anthropic says more than 70 agent platforms have integrated it since it was published as a standard in December 2025, including ChatGPT, Cursor, GitHub Copilot and VS Code. The same post says 1 million skills have been published and nearly 280 million installed.
What is being handed over: the standard, its name and logo, the GitHub organization and the agentskills.io domain. Engineers from Anthropic, Google, OpenAI and other companies will maintain it as individuals, with David Soria Parra and Mahesh Murag as lead maintainers.
What comes next: Anthropic has proposed starting with versioned releases, a formal process for proposing changes and security guidance for people who write and install skills. No dates were given.
Part of a larger batch: the AAIF announced four new projects the same day at its AGNTCon + MCPCon event in San Jose: Agent Skills (Anthropic), Open Responses (OpenAI), OpenEnv (Hugging Face) and OpenSandbox (Alibaba). That brings it to 10 hosted projects, up from 3 at launch in December 2025, alongside MCP and AGENTS.md.

What changes and what doesn't

What changes is who is in charge: the format no longer belongs to one company and moves to a foundation with neutral governance, as MCP did before it. If you were unsure whether writing skills was worth the time, this lowers the risk of being tied to one vendor.

The format itself doesn't change. A skill that works today keeps working, and Anthropic says anyone can keep using the standard, or contribute to it, without joining the AAIF. Nor does it change the fact that each tool decides what it supports: the specification itself flags the field that pre-approves tools as experimental and warns that support varies.

How a business can use it

Find what you keep explaining to the AI. How a quote is put together, your brand's tone, the format of a product page, the checklist before a page goes live (title, meta description, headings, alt text). That procedural knowledge is what, according to Anthropic, agents tend to lack for real business tasks.
Write it to the spec. A folder with a SKILL.md that opens with two required fields: "name" (up to 64 characters, lowercase letters, numbers and hyphens only, matching the folder name) and "description" (up to 1,024 characters, saying what it does and when to use it). Below that go the step-by-step instructions and an example or two.
Get the description right. At startup the agent reads only each skill's name and description, and loads the rest when a task calls for it. A vague description ("helps with the website") means the skill won't fire when it should; a specific one ("checks title, meta description and headings before a page is published") works better.
Keep it short. The spec recommends a SKILL.md under 500 lines, with longer material moved into separate files. Keep passwords, API keys and customer data out of the folder: it travels with the skill to every tool you install it in.
Check before installing someone else's skill. A skill can include scripts the agent runs. Before installing one from a public repository, check who published it and read what's in its "scripts" folder. The security guidance the maintainers have promised doesn't exist yet.

Sources

Updates: if the AAIF publishes the first versioned release or the security guidance for Agent Skills, it will be added here with a link.