command-creator

Create slash commands by defining Markdown files in .claude/commands/ with YAML frontmatter containing description and optional fields, then invoke them via /command-name. The frontmatter supports dynamic placeholders ($ARGUMENTS, $1, $2), file references with @, allowed-tools configuration, and optional Bash pre-execution for automation.

Command files use YAML frontmatter at the top to specify metadata like description, arguments, and tool constraints, followed by a Markdown body. Frontmatter can include disable-model-invocation, bash execution blocks, and file reference scoping; subdirectories under .claude/commands/ organize commands by project or context.

Yes. Commands support argument placeholders ($ARGUMENTS for all args, $1/$2 for specific positions) and file references using @ syntax. Bash pre-execution runs before the command, and allowed-tools restricts which tools the agent can invoke, giving you precise control over command behavior and scope.

Organize commands in .claude/commands/ subdirectories scoped to projects or users, and use symlink integration to share command definitions across environments. This lets teams maintain centralized command templates while keeping them available locally in each workspace.

The Skill includes YAML frontmatter validation and template scaffolding to ensure your command definitions are valid before deployment. This prevents syntax errors and makes it easier to bootstrap new commands with correct structure and optional fields.

Commands support allowed-tools to restrict tool access and disable-model-invocation to prevent model execution when needed. These model controls let you create commands that are purely Bash-based, tool-only, or fully model-integrated depending on your use case.