Claude Code Adds a Concise Output Style

Claude CodeView original changelog

Anthropic shipped a new built-in "Concise" output style for Claude Code that has Claude lead with the result and skip preamble and narration, while still completing work just as thoroughly as before. Developers can turn it on from Output style in /config, or by setting "outputStyle": "Concise" in settings.json, joining the existing Default, Explanatory, and Learning styles. The same release also fixed a prompt-caching bug that affected sessions routed through an LLM gateway or a custom base URL.

Key Takeaways

  • A fourth built-in output style has arrived: Concise joins Default, Explanatory, and Learning as a first-class, no-setup option in Claude Code.
  • Speed of reading, not speed of work, is the target: Anthropic says Concise skips preamble and narration but does the work "just as thoroughly," positioning it as a communication-style change rather than a capability tradeoff.
  • Two activation paths exist: developers can toggle it live from /config → Output style, or standardize it across a team by setting outputStyle: "Concise" in settings.json.
  • It targets experienced users: the style is aimed at developers who already trust Claude's judgment and want terse output for tight terminal or CI workflows.
  • The same release quietly fixed a caching bug: prompt caching for sessions behind an LLM gateway or custom base URL was broken and is now repaired, which should lower latency and cost for gateway users.
  • This extends a broader personalization trend: output styles have steadily expanded since their introduction, and Concise reflects continued investment in letting teams tune Claude Code's voice to their workflow.

A Fourth Built-In Output Style

Claude Code's output styles let developers reshape how Claude communicates without touching its core capabilities: file editing, command execution, and task tracking all stay the same regardless of which style is active. With version 2.1.237, Anthropic added a fourth built-in option, Concise, alongside the existing Default, Explanatory, and Learning styles.

Concise is built for developers who already trust Claude Code's judgment and want the fastest possible signal from a turn. Rather than narrating intermediate steps or explaining its reasoning as it goes, Claude leads with the outcome — results, diffs, test output — and keeps the surrounding text short. Anthropic is explicit that this is not a quality tradeoff: the style changes how much Claude says, not how thoroughly it works, and full detail remains available whenever a developer asks a follow-up question.

The style can be selected two ways: interactively through Output style in /config, or declaratively by setting "outputStyle": "Concise" in a project or user settings.json file, which makes it easy to standardize on for teams that prefer terse output in CI logs or shared terminals.

Prompt Caching Fix for Gateway Sessions

Version 2.1.237 also fixed a bug where prompt caching failed to work correctly for sessions using an LLM gateway or a custom base URL. Prompt caching is central to keeping long Claude Code sessions fast and affordable, so the fix should meaningfully reduce latency and token costs for teams routing traffic through a proxy or gateway layer rather than connecting to Anthropic's API directly.

Claude Code Adds a Concise Output Style | Yet Another Changelog