Frequently Asked Questions

Claude Code is Anthropic's terminal-based AI agent that can read, write, and edit files in your project. It runs directly from your command line, understands your entire codebase, and can autonomously complete complex multi-step tasks without constant supervision.

A Claude Skill is a SKILL.md file containing instructions that Claude loads automatically to customize its behavior for specific tasks. Skills tell Claude how to behave, what tools to use, and what rules to follow for a given context — like a configuration file for Claude's intelligence.

No. Claude Skills Hub is an independent community directory, not affiliated with Anthropic. We independently test, curate, and review skills to help developers find the best tools for their workflows. Always refer to Anthropic's official documentation for authoritative guidance.

Yes, completely free to browse and always will be. Browsing skills, reading guides, and accessing all directory features costs nothing. Listing your skill is also free — we never charge developers to have their work featured.

Our directory includes skills and resources for Claude Code, Cursor IDE, GitHub Copilot, VS Code, Windsurf, Gemini Code Assist, Cline, and OpenAI Codex. Many skills are cross-compatible, but always check the skill's documentation for which agents are officially supported.

Run npx skills add username/skill-name in your project root directory. This fetches the skill from GitHub and places it in .agents/skills/. Claude automatically loads all skills in that folder on the next session start. Each skill's listing page on this site shows the exact install command.

Skills are stored in .agents/skills/ inside your project directory. Each installed skill gets its own subfolder, for example .agents/skills/username/skill-name/. Claude scans this directory when it starts and automatically loads any SKILL.md files it finds there.

Yes, install as many skills as you need. Claude loads all skills it finds in .agents/skills/. There's no hard limit, though very large numbers of skills may impact context window usage. Group related skills and only install what you actively use for best performance.

Yes, we strongly recommend it. Committing .agents/ to Git shares your skill configuration with your entire team, ensures consistent Claude behavior across machines, and makes your project's AI setup reproducible. It's the same principle as committing .eslintrc or .prettierrc.

Delete the skill folder from .agents/skills/ and restart Claude. For example: rm -rf .agents/skills/username/skill-name. The skill will no longer be loaded on the next Claude session. You can also run npx skills remove username/skill-name if the CLI supports it.

Model Context Protocol (MCP) is an open standard created by Anthropic for giving Claude real capabilities like file system access, browser automation, database connections, and API integrations. Unlike skills (which modify Claude's behavior), MCP servers give Claude new tools to interact with external systems and services.

Run claude mcp add server-name -- npx -y @mcp/server-name in your terminal. Replace server-name with the server's identifier. Claude will prompt you to approve the connection and the tools it exposes. Always review what a server can access before connecting.

We vet every MCP server listed in our directory, but you should always review the source code yourself before connecting any server to Claude. MCP servers can have significant system access — only install from repositories you trust. Check our Security Audits page for our review process.

A skill is a SKILL.md file that provides instructions and context to Claude — it shapes how Claude thinks and behaves. A plugin is a code package that extends Claude's toolset with new commands, automations, or integrations. Skills modify cognition; plugins add functionality.

Most plugins are cross-platform and work on macOS, Linux, and Windows. Individual plugin pages in our directory list supported operating systems. Some plugins have system-specific dependencies — always check requirements before installing.

Browse all vetted plugins on our Plugins page. You can filter by platform, category, and compatibility. Each plugin listing includes an install command, description, and our review notes.

Re-run the install command for the plugin, or use your package manager's update command (e.g. npm update for npm-based plugins). If the plugin is installed as a skill, pull the latest version from the original repository and replace the files in .agents/skills/.

Use our Submit page. Fill in your skill details, install command, and contact email. We review every submission within 48 hours and notify you by email once it's approved or if we have questions.

Yes, always free. We believe a healthy AI skill ecosystem depends on open sharing, so we will never charge developers for listings. Approved skills stay listed permanently at no cost.

We reject skills with malicious or obfuscated code, private or unavailable repositories, inaccurate descriptions, duplicate listings, and skills that don't work as advertised. Read our Editorial Policy for the full review criteria.

No account needed. Browse, search, and read all skill documentation without registering. You only need to provide an email when submitting a skill — and that's purely for review notifications.

No. Claude Skills Hub is entirely free. All directory features, guides, and tools are available to everyone at no cost. We may explore optional sponsorship opportunities for featured listings in the future, but the core directory will always remain free.

Use our Contact page for general enquiries, partnership requests, or to report issues. For broken skill reports use the Report page. We aim to respond to all messages within 2 business days.