Anthropic donates Agent Skills standard to the Linux Foundation's Agentic AI Foundation

Anthropic announced it is donating Agent Skills to the Linux Foundation's Agentic AI Foundation, the same neutral home it chose for the Model Context Protocol. A skill is essentially a folder: a markdown file with instructions plus optional scripts and resources that an agent loads when relevant. Released in December 2025, the format has spread to more than 70 platforms, including competitors' products such as ChatGPT, GitHub Copilot, Cursor and VS Code.

Boris Cherny, Claude Code lead, explained the move: "Agent Skills started as an experiment in Claude Code. Many across the industry have come to adopt it, so we at Anthropic found it more fitting to donate it to the @AgenticAIFdn, next to MCP." Foundation governance means rivals can shape the spec without ceding control to Anthropic, which tends to accelerate adoption further.

Strategically, Anthropic is repeating the MCP playbook: ship a simple open format, let it become a de facto standard, then hand it to a neutral body. That gives Anthropic outsized influence over how agents are extended industry-wide, even when the agents run on other vendors' models. It also contrasts with proprietary plugin marketplaces such as xAI's new partner network.

Community reaction was muted and wry. One developer noted that "the widely accepted standard among over 70 platforms is simply a folder containing a markdown file." Security is the real watch item: one audit flagged only 0.33% of skills as malware risk, but common issues included piping downloads straight into shells. Neutral governance may bring signing and review requirements.