Anthropic is handing its AI agent playbook to an industry foundation, betting that shared standards will help software agents work across competing platforms.
The company announced Oct. 22 that it is donating Agent Skills to the Agentic AI Foundation (AAIF), a Linux Foundation organization it co-founded with Block and OpenAI. The transfer includes the standard’s name, logo, GitHub organization, and agentskills.io domain.
For IT teams, the potential benefit is reusing task instructions across compatible AI platforms instead of maintaining separate versions for each tool. Shared governance could help preserve that portability, although teams will still need to test how skills behave in each environment.
Anthropic developed Agent Skills while building Claude Code, after finding that general-purpose AI capabilities did not always translate into the procedural knowledge needed for practical business tasks. Each skill packages instructions in a folder containing a SKILL.md file, with optional scripts and reference materials that teach an agent how to perform a specific task.
Since Anthropic published the format as an open standard in December 2025, more than 70 agent platforms have integrated it, including ChatGPT, Cursor, GitHub Copilot, and VS Code. The company reports that developers have published 1 million skills, generating nearly 280 million installations.
Governance moves beyond one vendor
Anthropic says engineers from its own team, Google, OpenAI, and other companies will maintain the specification in their individual capacities. David Soria Parra and Mahesh Murag will serve as lead maintainers.
The arrangement follows Anthropic’s earlier donation of the Model Context Protocol (MCP) to the AAIF. Both projects address a central challenge for agentic AI: enabling systems and tools from different providers to work together.
Anthropic has proposed an initial roadmap covering versioned releases, a formal process for specification changes, and security guidance for people who create and install skills. Developers can continue using or contributing to the standard without joining the foundation.
Why the handover matters for enterprise AI
A common skill format could reduce the work required to maintain separate versions of the same business workflow for different AI platforms. Companies developing agents for software engineering, research, customer support, or internal operations could reuse instructions and supporting files across compatible environments.
That could make it easier for businesses to experiment with competing models without rebuilding every workflow from scratch. It also gives software vendors an incentive to support a shared format rather than require customers to adopt proprietary alternatives.
However, compatibility does not guarantee identical results. Skills may depend on particular software, network access, tools, or permissions that vary between agent environments. The foundation’s governance will also need to establish how changes are reviewed and how compatibility is maintained as the specification evolves.
More must-read AI coverage
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Why Data, Not Models, Determines AI Success
- The Rise of the AI-Native Factory: How Physical AI Is Transforming Manufacturing
What IT teams should check before reusing skills
Developers and businesses can continue using and contributing to Agent Skills without joining the foundation. For teams evaluating AI platforms, the practical question is how much of an existing workflow they can reuse—and which parts still require platform-specific changes.
Start with a low-risk task, such as formatting a report using sample data. Run the same skill with the same inputs across the platforms under consideration, then compare output quality, required tools, and approval prompts. A shared format can make instructions portable, but teams should verify that the receiving agent can access the dependencies and perform the steps the workflow requires.
Security needs a separate review. Skills can include executable scripts and instructions that direct an agent to use connected tools. Anthropic’s existing documentation warns that malicious skills can cause actions beyond their stated purpose. Before approving an unfamiliar skill, inspect its instructions and bundled code, identify external connections, and limit access to the files and systems needed for the task.
For production use, assign an owner to each approved skill, record the version tested, and review changes before deployment. Evaluate portability and security separately: a skill that runs successfully across several platforms still needs an assessment of what it can access and whether its outputs are reliable enough for the intended business task.
Read more: Explore how AI agents expose gaps in enterprise access controls and what security leaders should review before giving them access to business data.