Command Development

Create a Markdown file in .claude/commands/ for project commands or ~/.claude/commands/ for personal commands. The file content becomes the prompt Claude executes, and optional YAML frontmatter configures description, allowed-tools, model, and argument-hint.

Use $ARGUMENTS to capture all arguments as one string, or positional variables like $1, $2, $3 for individual arguments. Document expected arguments with the argument-hint frontmatter field so users see them in autocomplete.

Slash commands support description, allowed-tools, model (sonnet, opus, haiku), argument-hint, and disable-model-invocation. All fields are optional, and commands work with no frontmatter at all.

Yes. Use !`command` syntax to execute bash and include output, and @path/to/file to include file contents. Bash execution requires the command to allow the Bash tool via allowed-tools, ideally scoped like Bash(git:*).

Check that the file is in the correct commands directory, has a .md extension, and contains valid Markdown. Restart Claude Code after adding the file, and verify YAML frontmatter syntax is valid if used.

Plugin commands live in a plugin's commands/ directory and are discovered automatically when the plugin is installed. They can use ${CLAUDE_PLUGIN_ROOT} to reference plugin scripts, templates, and configuration portably across installations.