Free output style builder for Claude Code

An output style sets Claude's role, tone and response format for every reply in a session, so you stop repeating "keep it short" or "explain it like I'm not a developer" in each prompt. A custom style is a Markdown file: YAML front matter, then the instructions. Save it as ~/.claude/output-styles/<name>.md (all projects) or .claude/output-styles/<name>.md (one project), restart Claude Code, then switch with /output-style.

Everything runs in your browser. Nothing you type is sent anywhere. Field names and behavior follow the official output styles docs.

Start from a preset
Style name (shown in the /config picker)
Description (one line, shown in the picker)
Keep Claude Code's software engineering instructions (keep-coding-instructions: true). Tick this if Claude still writes code in this style; leave it off for writing, research or analysis. Role: who should Claude be?
Audience: who reads the replies?
Tone
Length
Format rules (one per line)
Always (one per line)
Never (one per line)
Example of a good reply (optional)
Save where?

Your output style file

Turn it on

Restart Claude Code so it reads the new file, then run this in the session (the command ignores case):

Or make it your default by adding this to . The settings value is case-sensitive and must match the style name exactly, or you silently get the Default style:

When an output style is the right tool (and when it isn't)

Use an output style for how every reply should sound and look for a whole session: voice, length, format, or a different role entirely (writer, analyst, tutor).
Use CLAUDE.md for facts Claude should know about your project or business. It stays loaded whichever style you pick.
Use a skill for instructions that apply to one kind of task, like a weekly client update.
Use a hook for anything that must happen every time without exception. A style is an instruction Claude follows, not a guarantee.
Subagents ignore your style. They run their own system prompt, so set tone inside the subagent's own file.
Built-ins first. If you only want shorter replies, try the built-in Concise style (/output-style concise) before writing your own.

Style not taking effect?

You edited the file during a running session: restart Claude Code.
The outputStyle value doesn't match the name exactly (for example client ready vs Client ready).
A misspelled front matter field is ignored without an error, and broken YAML loads the style under its file name with no fields set. Run claude --debug to see parse errors.
A project's settings file overrides the default in ~/.claude/settings.json.
Claude stopped coding carefully after you switched: your style dropped the engineering instructions. Set keep-coding-instructions: true.

Style not taking effect? See 12 reasons a Claude Code output style isn't working, and the fix for each.

More free tools: shape Claude's knowledge with the CLAUDE.md generator, add repeatable jobs with the SKILL.md builder, give specialists their own voice with the subagent builder, and enforce rules with the hooks generator. Not sure which feature fits? Read CLAUDE.md vs skills vs hooks: which goes where.

Made by Solo Stack Kits. If you'd rather start from finished examples, the free 3-skill sample has ready-made skills for client follow-ups and meeting notes, and the Solo Consultant Kit adds the full set.