Anthropic donated the Agent Skills standard (a folder with a SKILL.md plus scripts and reference files that an agent loads on demand) to the Agentic AI Foundation. The foundation already hosts MCP, A2A and AGENTS.md. Anthropic says more than 70 agent platforms support skills, including ChatGPT, Cursor, GitHub Copilot and VS Code, and that about 1 million skills have been published and nearly 280 million installed. The proposed roadmap starts with versioned releases, a formal proposal process and security guidance for people who write and install skills. AAIF also took in OpenAI's Open Responses, Hugging Face's OpenEnv and Alibaba's OpenSandbox.

Opinion: a skill is instructions plus code that an agent will follow and run, so a malicious skill is effectively a prompt injection you installed yourself. Until published guidance says otherwise, treat skills like any third-party package: pin them, review SKILL.md and bundled scripts, and load them only from sources you trust.