Claude Code Plugin Architect FAQs

Question 1

How does Claude Code discover new commands or skills?

Accepted Answer

Claude Code uses an auto-discovery mechanism that scans the commands/, agents/, and skills/ directories at the plugin root for compatible Markdown and configuration files.

Question 2

Can I define custom directories for my plugin components?

Accepted Answer

Yes, you can supplement the default discovery directories by specifying custom relative paths or arrays of paths within the plugin.json manifest.

Question 3

What is the difference between an Agent and a Skill in Claude Code?

Accepted Answer

Agents are specialized sub-entities with defined roles and expertise, while Skills are domain-specific guidance and instructions that Claude activates based on the task context.

Question 4

Why should I use ${CLAUDE_PLUGIN_ROOT} in my configuration?

Accepted Answer

Using the ${CLAUDE_PLUGIN_ROOT} environment variable ensures your plugin remains portable, as installation paths vary by operating system and user settings.

Question 5

What is the mandatory location for the Claude Code plugin manifest?

Accepted Answer

The manifest must be named plugin.json and must be located within a directory named .claude-plugin/ at the root of your plugin directory.

Claude Code Plugin Architect

Key Features

Use Cases

Claude Code Plugin Architect

Key Features

Use Cases