Command Development

Slash commands in Claude Code are custom prompts defined with YAML frontmatter that execute automated tasks. You configure dynamic arguments using `$ARGUMENTS` or positional parameters (`$1`, `$2`), optionally include file contents with `@path` syntax or execute bash commands with backticks, then invoke the command with a forward slash prefix to run your workflow instantly.

Yes, slash commands support dynamic arguments through `$ARGUMENTS` for flexible input or positional syntax like `$1` and `$2` for specific parameters. This lets you create reusable commands that adapt to different contexts—for example, a `/deploy` command that accepts environment and version inputs.

Use backtick syntax (`` !`command` ``) to execute bash commands directly within your command's prompt. The output integrates dynamically into the context, letting you reference git status, file metadata, or deployment scripts without manual copy-paste.

File references (`@path`) include file contents statically in your command, while bash execution (`` !`command` ``) runs a script and captures live output. Use file references for stable context like configuration or templates; use bash for dynamic data like system status or versioned artifacts.

Yes, slash commands support plugin-scoped and project-scoped deployment. Use `${CLAUDE_PLUGIN_ROOT}` for portable paths in plugins, and organize commands with integration points for agents, skills, and hooks to ensure consistency across deployments.

Slash commands include validation and error handling logic in their frontmatter configuration. Missing or invalid arguments trigger progressive disclosure—the command guides users to provide required inputs or reformulate their request before execution proceeds.