Anthropic moves Agent Skills to open governance

Anthropic is handing its Agent Skills standard over to the Agentic AI Foundation, moving the format for packaging procedural workflows under the Linux Foundation alongside the Model Context Protocol. At its core, the standard is intentionally lightweight: a folder containing a markdown instruction file, reference documents, and helper scripts that teach an agent how to complete a specific task. By keeping the interface simple, the format spread quickly across development environments, with Anthropic reporting nearly 280 million installs across tools like Cursor, GitHub Copilot, and ChatGPT.

Moving governance and the repository to a neutral body with maintainers from Google, OpenAI, and Anthropic is the right answer to platform lock-in. When teams build procedural knowledge for software development or operational routines, those workflows should not be trapped inside a single vendor's runtime. Standardizing how models receive contextual instructions means an organization can define a task routine once and run different agents against it as models evolve.

The immediate engineering hurdle now shifts to security and dependency management. Distributing executable scripts and instructions inside skill bundles creates an obvious software supply chain risk, especially as autonomous agents take on higher local permissions. Establishing strict versioning and provenance under neutral oversight will matter far more to enterprise adoption than the initial simplicity of the markdown file.