Plugin Structure

To structure a Claude Code plugin for auto-discovery, you must follow a standard directory layout that organizes commands, agents, skills, and hooks into specific component folders. This allows the system to automatically detect and load your components without manual registration.

The plugin.json manifest in Claude Code serves as the central configuration file that defines your plugin's architecture and component organization. It eliminates guesswork by specifying exactly how your commands, agents, and MCP servers are structured and deployed.

Yes, you can use portable path references when developing Claude Code plugins. This practice ensures your plugin architecture remains maintainable and deployable across different environments without breaking component auto-discovery mechanisms or file links.

The best way to organize commands and hooks in a Claude Code plugin is to follow the standard component organization guidelines. Separating these components into dedicated directories ensures maintainability and allows the auto-discovery mechanisms to function correctly during deployment.

Claude Code plugin architecture does not require external dependencies for scaffolding. You can create a new plugin with a single command by leveraging the standard directory structure and manifest configuration provided by the architecture guidelines.