Imagine writing a set of instructions once and having it work across Claude Code, Cursor, OpenAI Codex, Gemini CLI, VS Code, and a dozen more AI tools. That's exactly what Agent Skills deliver — and they're rapidly becoming the "npm packages" of AI-assisted development.
Announced by Anthropic on December 18, 2025, Agent Skills is an open standard that lets you package expertise, workflows, and automation into portable directories that any compatible AI tool can load and execute. As of March 2026, 16+ major AI tools have adopted the standard.
This isn't a proprietary feature locked to one vendor. It's an industry-wide shift in how we customize and extend AI tools. Think of it like this: if AI coding assistants are the new IDEs, Agent Skills are the extensions marketplace — except one extension works everywhere.
What Are Agent Skills?
Agent Skills is "a simple, open format for giving agents new capabilities and expertise." At the most basic level, a skill is just a directory containing a SKILL.md file. That markdown file tells an AI agent what the skill does, how to use it, and what rules to follow.
The Core Idea: Write Once, Works Everywhere
Before Agent Skills, customizing AI tools meant writing tool-specific configuration. Your Cursor rules didn't work in Claude Code. Your Codex setup didn't transfer to Gemini CLI.
Agent Skills changes that. One skill directory, one SKILL.md file, and it works across every tool that supports the standard — 16+ tools and growing.
Progressive Disclosure
When an AI agent starts up, it doesn't read every skill's full content. It only loads the name and description from the YAML frontmatter. The agent scans those short descriptions, and only when a skill is actually needed does it load the full content.
This means you can have dozens or even hundreds of skills installed without any performance penalty.
The Directory Structure
my-skill/
├── SKILL.md # Required — the skill definition
├── scripts/ # Optional — executable scripts
├── references/ # Optional — documentation, examples
└── assets/ # Optional — images, templates, configs
Which Tools Support Agent Skills?
The adoption has been remarkable. Within three months, 16+ major AI tools added support:
| Tool | Category | Developer |
|---|---|---|
| Claude Code | CLI Coding Agent | Anthropic |
| Claude (web/mobile) | AI Assistant | Anthropic |
| Cursor | AI Code Editor | Anysphere |
| OpenAI Codex | CLI Coding Agent | OpenAI |
| Gemini CLI | CLI Coding Agent | |
| Junie | IDE Agent | JetBrains |
| GitHub Copilot | AI Pair Programmer | GitHub |
| VS Code | Code Editor | Microsoft |
| OpenHands | Autonomous Agent | Open Source |
| Amp | AI Coding Agent | Sourcegraph |
| Goose | AI Agent | Block |
| Firebender | AI Coding Tool | Firebender |
| Letta | Agent Framework | Letta |
When Google, OpenAI, JetBrains, and GitHub all adopt the same standard, you know it's become the real deal.
The SKILL.md Format
Every skill starts with a SKILL.md file with two parts: YAML frontmatter and markdown content.
YAML Frontmatter Fields
| Field | Required | Description |
|---|---|---|
| name | Yes | Max 64 chars, lowercase + hyphens |
| description | Yes | Max 1024 chars — what agents read at startup |
| license | No | SPDX identifier (e.g., MIT) |
| compatibility | No | Which tools this skill targets |
| metadata | No | Custom key-value pairs (author, version, tags) |
| allowed-tools | No | Restrict which tools the skill can access |
Example SKILL.md
---
name: code-review
description: "Performs thorough code review with focus on security vulnerabilities, performance issues, and adherence to team coding standards."
license: MIT
compatibility:
- claude-code
- cursor
- codex
metadata:
author: your-team
version: 1.0.0
allowed-tools:
- read
- write
- exec
---
Creating Your First Skill: Step by Step
Let's build a "commit-message" skill that generates conventional commit messages.
Step 1: Create the Directory
mkdir -p my-project/.claude/skills/commit-message
Step 2: Write the SKILL.md
Create .claude/skills/commit-message/SKILL.md with name, description, and detailed instructions for analyzing git diffs, categorizing changes (feat/fix/refactor/docs/test/chore), and writing conventional commit messages.
Step 3: Add Optional Scripts
Skills can include executable scripts for validation, testing, or automation.
Step 4: Test It
Open your project in any supported tool and ask: "Write a commit message for my changes." The agent should detect and activate the skill automatically.
Claude Code Skills Deep Dive
Claude Code has the most mature implementation with advanced features:
Skill Locations and Priority
| Priority | Location | Path | Use Case |
|---|---|---|---|
| 1 (Highest) | Enterprise | Managed by org admin | Company-wide standards |
| 2 | Personal | ~/.claude/skills/ | Your workflow preferences |
| 3 | Project | .claude/skills/ | Project-specific standards |
| 4 (Lowest) | Plugin | Installed via skills.sh | Community skills |
Bundled Skills
- /simplify — Spawns 3 parallel review agents analyzing readability, performance, and correctness simultaneously
- /batch — Spawns 5-30 worktree agents for large-scale codebase changes
- /debug — Interactive debugging workflow that traces through code and proposes fixes
Skills Across Other Tools
Cursor
Cursor reads skills from .claude/skills/ in your project. Migrate your .cursorrules into Agent Skills format and they work in both Cursor AND every other supporting tool.
OpenAI Codex
Codex adopted Agent Skills as its customization format — a significant endorsement from Anthropic's biggest competitor.
Gemini CLI
Google optimized progressive disclosure for fast skill loading. Dozens of skills barely affect startup time.
VS Code
Skills placed in .claude/skills/ are automatically picked up by VS Code's AI assistant. Since VS Code is the most popular editor, this reaches the largest audience.
5 Practical Skill Recipes
- Code Review — Security, performance, and style checks with severity-rated output
- Test Generator — Generates comprehensive test suites for Jest, Vitest, Pytest, or Go
- Documentation Generator — Creates/updates README, API docs, and changelogs from code
- Security Audit — OWASP Top 10 analysis with dependency scanning and secrets detection
- Codebase Onboarding — Helps new developers understand unfamiliar codebases quickly
Skills Management with skills.sh
The community has built skills.sh — think of it as "npm for agent skills." You can discover, install, and manage skills from a central registry.
Security: 341 Malicious Skills Discovered
With any open ecosystem comes security risks. As of early 2026, 341 malicious skills have been discovered. Protect yourself:
- Always set allowed-tools for community skills
- Review SKILL.md content before installing
- Use enterprise-level skills for sensitive projects
- Keep skills version-controlled
How Serenities AI Uses Agent Skills
At Serenities AI, we use Agent Skills to standardize workflows across our integrated platform. Since Serenities AI lets you connect your existing AI subscription (ChatGPT Plus, Claude Pro) instead of paying per-token API costs, skills become even more powerful — you can run complex multi-step skill workflows without worrying about token costs eating into your budget.
FAQ
Do Agent Skills work with all AI tools?
As of March 2026, 16+ tools support the standard, including Claude Code, Cursor, OpenAI Codex, Gemini CLI, VS Code, JetBrains Junie, and GitHub Copilot.
Where should I put my skills?
For project skills: .claude/skills/. For personal skills: ~/.claude/skills/. The directory name is part of the standard, not brand-specific.
Can skills run code?
Yes. Skills can include scripts in the scripts/ directory. Use the allowed-tools field to restrict what untrusted skills can access.
Are Agent Skills free?
The standard itself is open and free. Individual skills may have their own licenses.
Bottom Line
Agent Skills represent a fundamental shift in how we customize AI tools. Write once, works everywhere, with progressive disclosure keeping things fast. Whether you're a solo developer or managing a team of hundreds, skills give you portable, version-controlled AI customization that moves with you across tools.
The ecosystem is still young, but with 16+ tools already on board, this is the standard that's going to define how we extend AI assistants for years to come.
Originally published at serenitiesai.com
Top comments (0)