Anthropic Donates Agent Skills Standard to Help AI Tools Work Across Different Platforms

Artificial intelligence tools are becoming more capable of performing complicated tasks, but they do not always work together smoothly.

For example, a developer may create a useful workflow for one AI assistant only to discover that another assistant requires a different setup.

Anthropic, the company behind Claude AI, wants to make these workflows easier to share.

On October 22, 2026, Anthropic announced that it was donating Agent Skills to the Agentic AI Foundation, an initiative under the Linux Foundation.

Agent Skills is an open standard that allows developers to package instructions and supporting resources that teach AI agents how to perform specific tasks.

According to Anthropic, more than 70 AI agent platforms have already integrated the standard, including ChatGPT, GitHub Copilot, Cursor, and Visual Studio Code.

But what exactly are Agent Skills, and why are they becoming important?

What Is an AI Agent?

An AI agent is an artificial intelligence system designed to carry out tasks involving multiple steps.

For example, imagine asking an AI assistant to prepare a monthly sales report.

A basic chatbot might explain how to create the report.

An AI agent connected to suitable tools could potentially retrieve sales data, analyse it, create charts, and prepare a document.

The agent may need instructions about which files to use, how calculations should be performed, and what the final report should contain.

This is where Agent Skills can help.

What Are Agent Skills?

Agent Skills are reusable packages of instructions and supporting resources designed for AI agents.

A skill typically contains a file named SKILL.md.

This file describes the task, the relevant procedure, and instructions the AI agent should follow.

The skill folder may also contain scripts, examples, and reference materials.

For example, a company could create a skill explaining how to prepare a standard financial report.

The instructions might describe how to organise the information, calculate important figures, and format the final output.

An AI agent that supports the standard could load those instructions when performing the task.

This can reduce the need to explain the same procedure repeatedly.

How Do Agent Skills Work?

Imagine a developer wants an AI agent to analyse an Excel workbook.

The developer could create a skill containing instructions such as:

• Check the workbook’s structure.
• Identify the relevant columns.
• Remove duplicate records where appropriate.
• Calculate summary statistics.
• Generate a report explaining the findings.

The instructions could be saved in a SKILL.md file.

Additional scripts could help the agent perform calculations.

When the skill is loaded by a compatible AI system, the agent can use the provided instructions to guide its work.

However, a skill does not automatically give the AI access to private files or business systems.

The agent must still have the necessary tools and permissions.

Why Is Anthropic Donating Agent Skills?

Anthropic originally developed the skills approach while working on Claude Code, its AI-powered software development tool.

The company found that general AI knowledge was not always enough for complicated professional workflows.

For example, an AI assistant might understand programming but lack the specific procedures required by a company’s development team.

Skills helped provide those instructions.

Anthropic published Agent Skills as an open standard in December 2025.

Since then, developers have created skills for a wide range of activities.

Anthropic now wants the standard to be maintained by a broader community rather than controlled by one company.

The Agentic AI Foundation will provide a neutral organisation for maintaining and developing it.

What Is the Agentic AI Foundation?

The Agentic AI Foundation, or AAIF, operates under the Linux Foundation.

The Linux Foundation supports collaborative technology projects, including open-source software and technical standards.

The foundation provides a framework through which different companies and developers can contribute to shared technologies.

Anthropic says the AAIF was established with participation from Anthropic, Block, and OpenAI.

The foundation also hosts the Model Context Protocol, commonly known as MCP.

By donating Agent Skills, Anthropic is helping move the standard towards community governance.

How Popular Are Agent Skills?

Anthropic reports that more than 70 AI platforms have integrated Agent Skills.

The company also reports that approximately one million skills have been published and nearly 280 million installations have occurred.

These figures are based on Anthropic’s reporting.

An installation does not necessarily mean a unique person has installed or actively uses a skill.

Nevertheless, the numbers suggest substantial developer interest in reusable AI workflows.

Which AI Tools Support Agent Skills?

Anthropic identifies several platforms that have integrated the standard.

These include ChatGPT, Cursor, GitHub Copilot, and Visual Studio Code.

Support may differ between platforms.

For example, one tool may automatically discover relevant skills, while another may require additional configuration.

Developers should consult the documentation for their specific AI environment.

What Is the Difference Between Agent Skills and MCP?

Agent Skills and the Model Context Protocol solve different problems.

Agent Skills provide reusable instructions and procedures that help an AI agent understand how to perform a task.

MCP provides a standard way for AI applications to connect with external tools and sources of information.

For example, an MCP connection might allow an authorised AI assistant to retrieve records from a database.

An Agent Skill could explain how those records should be analysed and presented.

The technologies can complement each other.

However, they are not interchangeable.

Why Does an Open Standard Matter?

An open standard is a publicly documented set of rules that different developers can implement.

Without shared standards, developers may need to rebuild similar features separately for each platform.

For example, a business might use one AI assistant for programming and another for document preparation.

If both support a common skills format, the business may be able to reuse some instructions across the tools.

This could reduce duplication and encourage collaboration.

However, compatibility is not always perfect.

Different AI agents may interpret instructions differently or support different features.

Are Agent Skills Safe?

Agent Skills can contain instructions and scripts, so security is an important consideration.

A malicious skill might attempt to persuade an AI agent to reveal sensitive information or perform unauthorised actions.

Developers should therefore examine skills before installing them, particularly when they come from unknown sources.

AI platforms also need appropriate permission systems and safeguards.

Anthropic says its proposed roadmap includes guidance on securely creating and installing skills.

Other priorities include versioned releases and a formal process for proposing changes to the standard.

What Does This Mean for Beginners?

Beginners interested in AI agents can start by learning how structured instructions work.

For example, someone who regularly prepares reports could document their workflow in clear steps.

Those steps might later become the foundation of an Agent Skill.

Programming knowledge can help when skills require scripts or more complicated processing.

However, some skills may consist mainly of instructions and reference material.

Understanding Markdown, files, APIs, and basic programming concepts can be useful.

QueryFun Take: Reusable AI Skills Could Make Automation Easier

The donation of Agent Skills reflects a growing effort to make AI tools more interoperable.

Instead of teaching every AI assistant a procedure from the beginning, developers may increasingly reuse structured instructions.

This could help businesses standardise tasks and reduce repeated configuration work.

However, sharing instructions between AI systems does not guarantee identical results.

Developers must still test workflows, protect sensitive information, and verify outputs.

For students interested in AI development, Agent Skills provides a practical introduction to automation, software standards, and AI agent design.

Frequently Asked Questions

What are Agent Skills?

Agent Skills are reusable packages containing instructions, scripts, and reference materials that help AI agents perform specific tasks.

What is a SKILL.md file?

It is a Markdown file containing instructions that describe how an AI agent should perform a particular task.

Has Anthropic donated Agent Skills?

Yes. Anthropic announced the donation to the Agentic AI Foundation on October 22, 2026.

Which AI platforms support Agent Skills?

Anthropic reports integrations with more than 70 platforms, including ChatGPT, Cursor, GitHub Copilot, and Visual Studio Code.

Are Agent Skills the same as MCP?

No. Skills describe procedures, while MCP provides a standard for connecting AI applications to tools and data sources.

Are Agent Skills free to use?

The specification is an open standard, but individual tools, hosting services, or commercial skills may have separate costs or conditions.

Can beginners create Agent Skills?

Yes. Beginners can start with clear written instructions and gradually add scripts or reference materials.

Primary sources:

Anthropic: Donating Agent Skills to the Agentic AI Foundation

Agent Skills Official Specification