On October 22, 2026 Anthropic donated the open Agent Skills standard to the Agentic AI Foundation (AAIF) at the Linux Foundation. A skill is a folder with a SKILL.md file that an agent loads for a specific task, and more than 70 platforms already support the format, including ChatGPT, Cursor, GitHub Copilot and VS Code. Below: what was handed over, what it changes for people learning to work with AI agents, and how to build your first Claude Code skill from the terminal in about 10 minutes.
The Terminal Agent for Absolute Beginners course has 6 lessons, about 76 minutes, from the phone chat to an agent that sorts your files and PDFs. The first 2 lessons are free, and signing up gives 7 days of full access with no card.
Facts checked on November 8, 2026 against Anthropic's announcement, the Linux Foundation press release and the Claude Code skills documentation.
Anthropic gave the foundation the Agent Skills name and logo, the standard's GitHub organization and the agentskills.io domain. The AAIF runs under the Linux Foundation; Anthropic, Block and OpenAI co-founded it in December 2025, and the Model Context Protocol (MCP) moved there earlier. Engineers from Anthropic, Google, OpenAI and other companies will maintain the standard as individuals, with David Soria Parra and Mahesh Murag as Lead Maintainers. You can keep using the format and contribute to it without joining the foundation.
The news came at AGNTCon + MCPCon North America in San Jose (October 22–23). The same day the foundation took in three more projects: Open Responses from OpenAI, OpenEnv from Hugging Face and OpenSandbox from Alibaba. The AAIF now hosts 10 projects, up from three at its launch in December 2025.
| Item | Value |
|---|---|
| Skills launched by Anthropic | October 2025 |
| Open standard published | December 2025 |
| Platforms that support it | more than 70, including ChatGPT, Cursor, GitHub Copilot and VS Code |
| Skills published | 1 million |
| Installs | nearly 280 million |
| Proposed first roadmap | versioned releases, a formal proposal process, security guidance for people who write and install skills |
The practical takeaway: an instruction you keep pasting into a chat can become a skill once and travel with you between agents. One format covers dozens of products, and a neutral foundation now governs it, so the time you put into your own skills is tied to no single vendor.
.claude/commands/deploy.md and a skill at .claude/skills/deploy/SKILL.md both create /deploy, and existing command files keep working.name, description, license, compatibility, metadata, allowed-tools. Claude Code extensions, such as injecting a command's output into the skill text, work only there.If the agent is not installed yet, start with our guide on how to run Claude Code in terminal: install, first session, exit and resume. The example below is a skill that turns a week of notes into a short report.
mkdir -p ~/.claude/skills/weekly-report
SKILL.md file in that folder: a header between --- lines, then plain-language instructions.---
name: weekly-report
description: Builds a weekly report from the notes in the notes folder. Use when asked for the week's results or a report for a manager.
---
1. Read every file in the notes folder from the last 7 days.
2. Group the work by project, 1-3 bullet points per project.
3. List what slipped and why in a separate block.
4. Keep it to one page, with no evaluative words.
name is up to 64 characters of lowercase letters, digits and hyphens, with no double hyphen or hyphen at either end, and it matches the folder name. Keep SKILL.md under 500 lines and move long reference material into separate files next to it.claude./weekly-report.To share a skill with your team, put it in the repository at .claude/skills/weekly-report/SKILL.md and commit it. Custom commands and keeping an agent in bounds are covered in the slash commands lesson of the developer course.
A skill can carry scripts and pre-approved tools. The Claude Code documentation advises reviewing the allowed-tools field of skills checked into a repository before you run Claude Code there, because a skill can grant itself broad tool access. Security guidance for people who write and install skills is part of the roadmap Anthropic proposed for the standard. Until it lands, install skills from sources you know and read the whole SKILL.md before the first run.
| Course | Who it is for and what is inside |
|---|---|
| Terminal Agent for Absolute Beginners 6 lessons, 76 min | No coding: starting the agent, first tasks with files, what to trust in its answers |
| Claude Code: Agentic Development on Real Projects 10 lessons, 240 min | CLAUDE.md, your own slash commands, plans with checkpoints, MCP, IDE and CI |
The first 2 lessons of every course are free; sign-up opens the whole catalog for 7 days with no card, then Pro is $9 a month or $79 a year (pricing). Your Claude plan is paid separately to Anthropic.
An open format for giving AI agents skills. A skill is a folder with a SKILL.md file: a header with a name and a description, instructions below it, and optional scripts and reference files next to it. The agent sees the list of skills and loads one when a task matches its description.
Anthropic donated the Agent Skills standard to the Agentic AI Foundation at the Linux Foundation: the name, the logo, the GitHub organization and the agentskills.io domain. Engineers from Anthropic, Google, OpenAI and other companies now maintain it, and you can use it without joining the foundation.
A personal skill goes in ~/.claude/skills/skill-name/SKILL.md and works in all your projects on that computer. A project skill lives in .claude/skills/skill-name/SKILL.md inside the repository; commit it and your whole team gets it.
Yes, as long as the SKILL.md header uses only the standard fields: name, description, license, compatibility, metadata and allowed-tools. Features that Claude Code adds on top, such as injecting a command's output into the skill text, work only in Claude Code.
Start with the free lessons of Terminal Agent for Absolute Beginners and write your first skill after the lesson on real tasks. Sign up for 7 days of full access with no card.