Anthropic Transfers Agent Skills to Linux Foundation
The open standard for AI agent task automation moves to the Agentic AI Foundation to foster industry-wide development.
Updated on Oct. 22, 2026 in Artificial Intelligence
Live Poll
Do you trust the security of automated AI tools downloaded to your computer?
Anthropic has officially transferred control of its Agent Skills project and the associated SKILL.md file format to the Linux Foundation’s Agentic AI Foundation. This transition moves the popular technical standard under a neutral governing body as it reaches 1 million published skills.
Why it matters
The project was originally developed because AI models like Claude Code lacked the specific procedural knowledge required for complex business tasks. By standardizing these instructions, the platform aims to create a consistent framework for agent-based automation across different systems.
The SKILL.md format consists of a folder containing instruction files and scripts that agents load at startup, with full instructions read only upon task matching. Currently, more than 70 agent platforms have integrated this standard.
The players
Anthropic
An artificial intelligence safety and research company that develops the Claude series of large language models.
Agentic AI Foundation
A Linux Foundation initiative co-founded by Anthropic, Block, and OpenAI to oversee open standards for AI agent interoperability.
David Soria Parra
A lead maintainer for the Agent Skills project who works alongside the open-source community to evolve the SKILL.md format.
JFrog
A software supply chain company that provides security analysis and identified nearly 1,000 malicious payloads in existing agent skills.
The details
The transfer includes the agentskills.io domain, the project GitHub organization, and all associated branding. The project remains maintained by individual engineers from Anthropic, Google, and OpenAI, including leads David Soria Parra and Mahesh Murag.
Timeline
October 2025: The Agent Skills project first shipped.
December 2025: Anthropic published the format as an open standard and co-founded the AAIF with Block and OpenAI.
October 22, 2026: Anthropic transferred the Agent Skills project to the Linux Foundation.
The Tech Race
The transfer to the Agentic AI Foundation signals a shift toward vendor-neutral governance for critical AI automation standards. This move mirrors the evolution of other open-source frameworks where cross-industry collaboration replaces proprietary control to ensure long-term stability.
Users of AI agent platforms will likely see more consistent task execution as the standardized SKILL.md format gains wider adoption. Developers can expect more robust security guidance as the project implements new versioning and formal proposal processes.
The takeaway
Standardizing how AI agents access procedural knowledge is critical for reducing errors in complex business environments. As the volume of available skills grows into the millions, vetting these tools for security payloads remains a vital responsibility for the platform maintainers.
Further reading
Learn more about the evolving ecosystem of open-source automation in our Artificial Intelligence section.
Live Poll
Do you trust the security of automated AI tools downloaded to your computer?
