claude-commands

Custom slash commands in Claude Code are defined using markdown-based command files with static text, dynamic arguments (`{{arg}}`), and conditional sections (`{{#if}}`). Place command files in your plugin or user command directory, register them in `plugin.json` with kebab-case naming, and Claude Code loads them automatically for instant access.

Yes. Embed shell commands like `{{shell:git diff --staged}}` directly in your command definition to automate Git workflows. For example, create a `/gcm` command that analyzes staged changes and generates conventional commit messages, eliminating manual composition.

Organize commands using directory-based loading and `plugin.json` registration. Use clear kebab-case naming conventions, document command arguments for discoverability, and structure related commands within their respective plugin roots to maintain clean separation and cross-plugin workflows.

Define arguments using `{{arg}}` for positional parameters and `{{named_arg}}` for named parameters in your command file. Claude Code parses both single and multiple argument types, allowing you to build flexible commands that accept user input dynamically.

Yes. Include file references and shell commands within a single command definition using `{{shell:...}}` syntax and file inclusion patterns. This enables multi-step workflows where one command orchestrates complex sequences across templating, file access, and command execution.

Debug command execution by verifying registration in `plugin.json`, checking directory placement against plugin roots and user directories, validating argument syntax (`{{arg}}` format), and ensuring kebab-case naming. Claude Code's directory-based loading and cross-plugin workflows surface configuration issues during execution.