Output Styles

Output Styles change how Claude Code responds, not what it knows: they modify its system prompt to set the agent's role, tone, and output format. Claude Code ships with four built-in styles — Default (the standard software-engineering prompt), Proactive (acts immediately and favors action over planning), Explanatory (adds educational "Insights" about implementation choices while it works), and Learning (a collaborative mode that adds TODO(human) markers and asks you to write small pieces yourself). They can also adapt Claude Code for non-coding work like writing or data analysis.

You pick a style by running /config and choosing Output style; your selection is saved to .claude/settings.local.json, and you can also set the outputStyle field directly in a settings file. Because the style is read into the system prompt once at session start, changes take effect after /clear or a new session. (The standalone /output-style command was deprecated and later removed in favor of /config.)

Custom output styles are Markdown files — frontmatter for metadata, then instructions appended to the system prompt — stored at the user level (~/.claude/output-styles) or project level (.claude/output-styles), so they can be versioned and shared like slash commands and Agent Skills. By default a custom style drops Claude Code's built-in coding instructions unless you set keep-coding-instructions: true. Because an output style edits the system prompt directly, it differs from CLAUDE.md (added as a message after the prompt) and from subagents (separately scoped helpers). For exact configuration details, see the official Claude Code docs.