Skip to main content

Output Styles

Official guide for adapting Claude Code for uses beyond software engineering through output styles that modify the system prompt. Source: https://code.claude.com/docs/en/output-styles.md Added: 2026-01-29

Overview​

Output styles allow you to use Claude Code as any type of agent while keeping its core capabilities (running scripts, reading/writing files, tracking TODOs).

Built-in Output Styles​

StyleDescription
DefaultStandard system prompt for software engineering tasks
ExplanatoryProvides educational "Insights" between helping with tasks. Helps understand implementation choices and codebase patterns
LearningCollaborative learn-by-doing mode. Shares "Insights" and asks you to contribute code with TODO(human) markers

How Output Styles Work​

  • All styles exclude instructions for efficient output (concise responses)
  • Custom styles exclude coding instructions unless keep-coding-instructions is true
  • Custom instructions added to end of system prompt
  • Reminders injected during conversation to maintain adherence

Changing Output Style​

/output-style                    # Menu selection
/output-style explanatory        # Direct switch

Saved in .claude/settings.local.json at local project level. Can also edit outputStyle field in settings directly.

Creating Custom Output Styles​

Custom output styles are Markdown files with frontmatter:

---
name: My Custom Style
description: Brief description for UI display
---

# Custom Style Instructions

You are an interactive CLI tool that helps users with...

## Specific Behaviors
[Define behavior...]

Storage Locations​

  • User level: ~/.claude/output-styles/
  • Project level: .claude/output-styles/

Frontmatter​

FieldPurposeDefault
nameDisplay nameInherits from filename
descriptionUI descriptionNone
keep-coding-instructionsKeep coding-related system promptfalse

Comparisons​

Output Styles vs CLAUDE.md vs --append-system-prompt​

  • Output Styles: Turn off default SE system prompt entirely
  • CLAUDE.md: Added as user message following default prompt
  • --append-system-prompt: Appended to system prompt (doesn't replace)

Output Styles vs Agents​

  • Output Styles: Affect main agent loop, only modify system prompt
  • Agents: Invoked for specific tasks, can include model, tools, context settings

Output Styles vs Skills​

  • Output Styles: Modify how Claude responds (formatting, tone). Always active once selected
  • Skills: Task-specific prompts invoked with /skill-name or auto-loaded when relevant