On 22 October 2026, Anthropic announced it is donating Agent Skills to the Agentic AI Foundation (AAIF), which sits under the Linux Foundation. Agent Skills is an open standard for giving AI agents specialized knowledge and workflows.
What happened
A skill is a folder with a SKILL.md file. The file teaches an agent one task or domain. It can also bundle scripts and reference material. Anthropic built the idea while working on Claude Code, which handled general knowledge work well but lacked the procedural knowledge that real business tasks need.
The donation covers the standard itself, the Agent Skills name and logo, the GitHub organization and the agentskills.io domain. Anthropic co-founded AAIF with Block and OpenAI in 2025. It donated the Model Context Protocol to the same foundation.
Anthropic reports that the standard was published in December 2025 and that more than 70 agent platforms have integrated it. It lists ChatGPT, Cursor, GitHub Copilot and VS Code among them. It also reports 1 million published skills and nearly 280 million installs. Engineers from Anthropic, Google, OpenAI and other companies will maintain the standard as individuals.
The proposed roadmap covers versioned releases, a formal proposal process and security guidance for people who write and install skills.
What it means for people who check data
A skill is a written procedure. That makes it something you can read, review and version, which is useful when you need to explain how an AI result was produced. Our how it works page describes the same idea for human review: a repeatable process beats a one-off judgment.
There is also a caution. A skill can include scripts, and an installed skill shapes what an agent does with your data. Treat third-party skills as unreviewed code until someone on your team has read them. The security guidance on the roadmap is worth watching for that reason.
Practical steps
- Keep a list of which skills your agents use, and who approved each one.
- Re-run data verification on outputs when a skill changes.
- Record skill versions next to the results they produced, so a figure can be traced back through the trust index style of evidence scoring.
Open governance does not make a skill correct. It makes the standard easier to inspect. The checking still has to be done.