What Is the Agent Skills Standard? A No-Code Guide

Ellie at Willa, Editorial lead, Willa · Published November 2, 2026 · Updated November 2, 2026

If you built a Claude Skill this year, you now own something more valuable than you did a month ago. On October 22, Anthropic donated the Agent Skills standard to the Agentic AI Foundation, a neutral body under the Linux Foundation. That sounds like news for lawyers and engineers. It is actually news for you, because it means the skill you wrote in plain English is no longer tied to one chatbot. More than 70 platforms, including ChatGPT, Cursor, GitHub Copilot, and VS Code, have adopted the same format.

Most of the coverage so far has been written by developers, for developers. This guide explains what the Agent Skills standard is, what the donation changes, and what to do about it if you are a non-technical person who builds with AI, in the time it takes to drink a coffee.

What is the Agent Skills standard?

The Agent Skills standard is a shared, open format for packaging instructions that an AI assistant can follow. A skill is simply a folder. Inside that folder sits a plain-text file called SKILL.md that teaches the AI a specific task, in your words: how you draft a newsletter, how your team formats meeting notes, how you qualify a sales lead. The folder can also hold reference files or scripts, but for most non-technical skills, the SKILL.md file does all the work.

Anthropic launched skills inside Claude in October 2025, then published the format as an open standard in December 2025 so that any AI tool could read the same files. Adoption was fast. According to Anthropic's announcement, more than 70 agent platforms have integrated Agent Skills, over 1 million skills have been published, and skills have been installed nearly 280 million times.

A useful way to think about it: Agent Skills are becoming the PDF of AI instructions. You write the file once, and it opens the same way everywhere.

Do Claude skills work in ChatGPT?

Yes. ChatGPT is one of the platforms Anthropic names as having adopted the standard, alongside Cursor, GitHub Copilot, and VS Code. A skill you wrote for Claude is, at its core, a folder of instructions in an open format, and any tool that supports the standard can load that same folder.

Two honest caveats. First, each platform decides how you add a skill, so the exact menus and setup steps differ from tool to tool. Check your tool's own documentation for "Agent Skills" or "SKILL.md" to see how it handles installation. Second, platforms vary in how deeply they support extras like bundled scripts. The plain-English instructions at the heart of your skill are the portable part, and that is the part most non-technical builders rely on anyway.

The practical takeaway: keep a copy of your skill folders somewhere you control, like a folder on your computer or in your cloud drive. That folder is now a portable work asset, the same way a resume or a portfolio is.

Why did Anthropic give the standard away?

The Agentic AI Foundation (AAIF) is a project under the Linux Foundation, the same nonprofit that stewards much of the open-source software the internet runs on. Anthropic co-founded the AAIF in 2025 with Block and OpenAI, and had already donated the Model Context Protocol (MCP), the standard that lets AI tools connect to apps like Slack and Google Drive.

The donation is complete: the standard itself, the Agent Skills name and logo, the GitHub organization, and the agentskills.io domain all move to the foundation. Going forward, engineers from Anthropic, Google, OpenAI, and other companies will maintain the standard as individuals, and the published roadmap includes versioned releases, a formal proposal process, and security guidance for people who write and install skills. Developer Simon Willison, who tracks this space closely, predicted this move back in December 2025 when Anthropic first called skills an open standard.

Why this matters to you: when one company owns a format, your work lives at that company's pleasure. When a neutral foundation owns it, with maintainers from competing companies, the format is much more likely to outlive any single product decision. Time you invest in writing skills is now a safer investment.

What is inside a skill, exactly?

Less than you might expect. A SKILL.md file has two parts: a short header with the skill's name and a one-line description, then the instructions themselves, written in ordinary markdown. Here is the shape of one:

---
name: weekly-update-draft
description: Drafts my Friday team update in my voice from my bullet points
---

When I give you raw bullet points for the week, turn them into
my Friday update. Lead with the single most important outcome.
Keep it under 200 words. Use plain language, no jargon.
End with next week's top priority, phrased as a commitment.

That is a real, working skill. No code, no setup wizard, no API keys. If you can write a clear note to a new teammate, you can write a SKILL.md file. The full specification lives at agentskills.io if you want to see every option, but you do not need most of them to start.

How do you build a portable skill without coding?

Here is the short version of the process, with the full walkthrough in our guide to creating a Claude Skill without coding.

1. Pick one task you repeat weekly. The best first skill is something you already re-explain to AI over and over: a report format, an email style, a review checklist.
2. Write the steps as if onboarding a new teammate. Include what good output looks like, what to avoid, and one or two real examples. Specifics beat adjectives: "under 200 words" works better than "keep it brief."
3. Ask your AI to turn it into a skill. In Claude, you can describe your process in a normal conversation and ask it to create the skill for you, including the SKILL.md file and folder structure.
4. Test it on real work, then tighten it. Run the skill on three real tasks. Where the output drifts, add a line to the instructions. Skills improve by editing, not by starting over.
5. Save the folder somewhere you control. Download a copy. This is the step most people skip, and it is the one that makes your skill portable across Claude, ChatGPT, and whatever you use next year.

What does this mean if you built Custom GPTs?

If you spent 2024 or 2025 building Custom GPTs, you already know how it feels when a platform-owned format reaches end of life. We covered that transition in Custom GPTs are retiring: how to migrate to plugins, and the lesson from it is the reason this announcement matters: instructions locked inside one vendor's product eventually move, change, or disappear on that vendor's schedule.

An open standard reduces that risk. If you still have instruction text from old Custom GPTs sitting in a doc somewhere, converting it into SKILL.md files is a genuinely good weekend project. The writing you did, the voice, the rules, the examples, carries over almost unchanged. You are not starting from scratch; you are moving your work into a format with many doors instead of one.

Key takeaways

• A skill is a folder with a plain-text SKILL.md instruction file. If you can write clear instructions in English, you can build one. No code required.
• As of October 22, 2026, the Agent Skills standard belongs to the Agentic AI Foundation under the Linux Foundation, not to Anthropic, which makes it vendor-neutral.
• More than 70 platforms, including ChatGPT, Cursor, GitHub Copilot, and VS Code, support the format, with 1 million skills published and nearly 280 million installs so far.
• Your existing Claude Skills keep working exactly as they did. Nothing changes for builders, except that your skills are now a portable asset worth backing up.
• Skills written in plain English are the most portable kind, so non-technical builders are arguably the best positioned for this shift.

Your skills are a career asset now

A year ago, "I built a few AI skills for my team" was a nice line in a standup. Now it describes portable, documented work product that moves with you between tools and between jobs. That is worth putting real care into, and it is a lot more fun to figure out with other women doing the same thing.

If you want to see how other non-technical builders structure their skills, or you want feedback on your first one, come to a Willa event. And if you are building a career with AI and want a community that shares what actually works, join Willa. We would love to have you.

• Agent Skills
• Claude Skills
• ChatGPT
• open standards
• no-code

Join Willa, the community where women learn AI together.

All posts