Claude Styles Explained: Tone, Length, and Preferences
Claude styles let you save preferences for how Claude writes—tone, length and format—so you do not re-explain them every session. Here is what styles are, how to use them and how to make them work across sessions and AI apps.
Updated 5 min read
Claude styles are saved sets of preferences for how Claude writes—its tone, sentence length, format choices and communication approach. Instead of re-typing "be concise, use plain language and avoid bullet points" at the start of every session, you save those instructions once and apply them with a click. Tools like fenbs extend this further: your approved preferences live outside the Claude app, so other connected AI apps can use the same rules too.
Understanding styles matters because Claude's default output is a compromise: moderately detailed, moderately formal, with a tendency toward lists and hedging. For most personal or professional work, that default needs adjustment. Styles are the mechanism for making that adjustment permanent.
What a Claude style actually contains
A style is a named collection of writing instructions. The Claude app ships with a few preset options—concise, explanatory, formal, casual—but you can create custom styles that describe exactly how you want Claude to write for a particular purpose.
A custom style typically includes:
- Tone and register: formal, conversational, direct, warm.
- Sentence and paragraph length: short and punchy, long and detailed, or a specific maximum.
- Format preferences: flowing prose, structured sections, numbered steps, or no bullet points unless asked.
- Vocabulary choices: plain language, technical terms, contractions, specific words to avoid.
- Response length: a word count target or an instruction to match the length of the question.
You can also include sample text that demonstrates the target style—a paragraph written the way you want. Claude matches patterns from examples as well as from instructions.
How to set up a style in Claude
In the Claude app, styles appear as a control in the conversation interface. You can select a preset or open the style editor to write a custom one. Give it a name that identifies the context—"Client emails," "Technical docs," "Weekly reports"—so you can pick the right one quickly.
Write each preference as a specific instruction rather than a general label. "Keep responses under 150 words unless I ask for more" is more useful than "be brief." Test the style on a real task before saving it: send Claude a message you would normally write, read the result and adjust until it matches what you wanted.
Styles, projects and where instructions live
Styles in the Claude app apply within the app. Claude Projects let you add a persistent instruction set to a project folder that loads at the start of every conversation there. The two work together: a project instruction can set the topic context while a style sets the communication preferences.
The limitation is portability. Styles you create in Claude do not automatically transfer to other AI apps. If you use Cursor, Codex CLI or ChatGPT alongside Claude, each one starts without your preferences unless you supply them separately. Claude Projects add a persistent instruction layer on top of styles; combining both gives you topic context and communication preferences in one place.
Keep your preferences where every connected AI can read them
Store your approved writing rules in fenbs Skills and prompts, and your standing preferences in Decisions and rules. Any AI app you connect reads them at the start of a session over MCP—so the same communication preferences work in Claude, Cursor, ChatGPT or any other connected assistant without being re-entered in each one.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
Styles for different kinds of work
Most people need more than one style. A client-facing email style is different from the one you want for internal notes, which is different again from how you want Claude to explain a technical concept. Create a separate style for each major context rather than trying to write one that covers everything.
A practical starting set for many professionals: one for external communication (formal, careful, clear), one for internal work (faster, more direct, assumptions allowed) and one for writing that will be edited (show a clear structure, leave opinions labeled as options). Three focused styles are easier to maintain and faster to select than one overly complex one.
The concise style: when and how to use it
The built-in concise style instructs Claude to answer directly and briefly. It is useful for quick lookups, code explanations and anything where a two-sentence answer is better than a page. The trade-off is that nuance gets cut: a concise style on a technical question may leave out the edge cases you needed.
A custom concise style can be more precise than the preset. Specify the maximum word count per response, when Claude is allowed to elaborate (if asked, or if a safety issue arises) and whether bullet points are acceptable. That avoids the situation where "concise" produces a terse list when you wanted a short paragraph.
What to do if your styles stop working
Styles occasionally reset, fail to apply or appear unavailable after an app update. The fix steps and what causes styles to go missing are covered in claude styles gone. The short version: check whether a style is still selected, whether the session has a conflicting instruction, and whether the style was saved to the account or to a specific device.
How to do this with fenbs
- Create a free account and add a project for your main area of work.
- Write your communication preferences—tone, length, format, vocabulary—as Decisions and rules entries. One preference per entry, with the reason it matters.
- Save your reusable prompts and sample text in Skills and prompts. Name each one by context, such as "Client email style" or "Technical explanation style."
- Connect Claude to fenbs from Integrations. Ask it to read the project context at the start of each session so your preferences load automatically.
- When you update a preference after testing, change it in fenbs and every connected assistant picks it up from the next session.