generating-output-styles

Output styles let you customize Claude Code's system prompt via markdown files with YAML frontmatter. Create a file in `.claude/output-styles/` or `~/.claude/output-styles/`, then activate it with `/output-style` to tailor responses for writing, teaching, or brainstorming tasks.

Output style files use YAML frontmatter to define metadata like name and description, followed by the custom system prompt content. Valid frontmatter must include proper naming constraints, keep-coding-instructions flags, and be placed in the correct directory for project or user-level activation.

Yes, output styles support both placement options: `.claude/output-styles/` for project-specific styles and `~/.claude/output-styles/` for user-level styles. Project-level styles override user-level ones when both exist in the same workflow.

After creating your output style file with proper frontmatter and naming conventions, run the `/output-style` command in Claude Code to view available styles and activate your custom one. This lets you verify the style works before using it in tasks.

Output style files must have valid YAML frontmatter, follow naming and description constraints, include appropriate keep-coding-instructions flags, and be placed in `.claude/output-styles/` or `~/.claude/output-styles/`. Invalid frontmatter or placement will prevent activation.

Yes, output styles are designed specifically to customize Claude Code for non-engineering tasks by changing system prompts. This allows you to adjust response format, tone, and behavior for writing, teaching, analysis, and other non-code workflows.