# Should I use Ownhand or a Claude style to write like me?

Anthropic is folding Claude styles into skills, and a custom style built from your writing samples moves over as a skill you switch on in the Skills section of Customize; that free skill is still the right pick for writing you do inside Claude, while Ownhand fits text that goes out under your name from any agent, fitted to one of 12 occasions or to the thread you are answering, and improved by the exact text you send.

By Jordan Gibbs, founder of Ownhand at Thalient Labs. Updated 4 October 2026.

Claude facts come from Anthropic's help center as of 4 October 2026. Two of those pages have since been taken down, so I link archived copies of them. I run Ownhand, so check my reading against the links.

## What changed with Claude styles

Styles lived in a Use style menu in the Claude chat box. You could pick a preset such as Concise or Formal, or build a custom style. The custom route took a writing sample, uploaded as a PDF, DOC or TXT file or pasted in. Claude read it and wrote a style: a block of instructions you could preview, rename and edit by hand.

In a help center notice last updated 1 June 2026, Anthropic said styles are becoming skills. Custom styles get migrated into skills automatically, switched off until you turn them on. The Concise, Explanatory and Formal presets go away. Learning survives as a skill you add from the directory. Once the migration finishes, the styles menu disappears from Claude.

That notice is gone now too. The old Styles URL returns a not found page. Anthropic's personalization guide, updated 16 September 2026, lists three ways to shape Claude: Instructions for Claude, project instructions and skills. It says you can use skills to apply communication patterns based on your own writing.

So the real comparison today is a Claude skill, possibly one that started life as your style, against Ownhand.

## Side by side

|  | Claude skill (or a migrated style) | Ownhand |
| --- | --- | --- |
| What it is | A folder with a `skill.md` file of instructions, plus any reference files | An MCP server that rewrites a finished draft from your Hand |
| Where the voice comes from | Instructions, often written by Claude from your samples, plus examples you add | 5 to 10 things you sent, a line or two on how you write, and rules learned later |
| When it runs | When Claude judges it relevant, or when you type its slash command | When your agent calls `write` on text you will send as yourself |
| Slack DM vs status update | One set of instructions unless you write both cases out | 12 occasions, from `chat_dm` to `status_update`, plus up to 20 of your own |
| Replies | Whatever Claude picks up from the pasted thread | Measures casing, length, greetings and punctuation in the thread and matches them |
| Changes over time | When you edit the file | After each message your agent reports as sent |
| Where it works | Claude apps, Claude Code and the API, plus other tools that adopt the Agent Skills standard | Any MCP client: Claude, ChatGPT, Cursor, Codex and more |
| Cost | Included in every Claude plan, with code execution on | $1 free, then about half a cent per rewrite |

## When a skill is all you need

If most of your writing happens inside a Claude chat, keep the skill. Drafting a long memo, a blog post or a grant section with Claude and editing it there is exactly what a style was for. The free option wins here. You see every line before it leaves, so a missed habit costs you a quick edit, and you pay nothing extra.

It also wins if you write in one register. A consultant who sends polished client emails all day can describe that voice once and be done. Ownhand earns its price when the same person writes in five places that each sound different.

If your style came over from the migration, open it before you rely on it. It holds the instructions Claude wrote from your samples. If they read like a list of adjectives, add three to five real messages as examples, label where each one went, and rewrite the description so Claude knows when to load it. The help center caps that description at 200 characters. [How to make Claude write like you](https://ownhand.dev/answers/make-claude-write-like-me) walks through building a skill from scratch.

## Where a skill runs out

A skill is a set of instructions, and instructions get followed loosely. Nothing checks the output against them. If the draft opens with a stock pleasantry you banned, you catch it or the reader does.

It also stays still. Say you cut every sign-off from Claude's drafts for a month. The skill never hears about it, because the edits happen in Gmail or Slack after you copy the text out. You learn your own habits by noticing them, then go back and type them into the file.

And it lives in Claude. Skills follow an open standard, so a well-built one can travel to other tools that adopt it, but you still carry the file around and keep the copies in sync.

## What Ownhand does in that gap

Ownhand sits at the last step, after the agent has a draft and before you send it. The agent calls `write` with the draft and the occasion. Ownhand rewrites it from your Hand and checks the result for AI-favored words, stock sentence patterns, the occasion's length and shape, and every name, number, date and link from the draft. A failed check goes back to the model for up to two more tries. Anything still wrong comes back flagged, so you see it before you approve.

After you send, the agent reports the exact text that went out. The difference between draft and sent text turns into rules you can read, pin or reject in the dashboard. The full loop is in [how Ownhand learns from what you send](https://ownhand.dev/answers/how-ownhand-learns-from-edits).

Small habits are where samples beat a summary. Here is one stiff request rewritten for two writers:

These examples are real output from the live API. Each AI draft was rewritten by a demo Hand: a fictional writer (an engineer, a founder, an ops lead, a designer or a consultant) built from a handful of samples.

**AI draft:** "Would you be available for a quick sync sometime this week? I'm flexible and happy to work around your schedule. Tuesday after 2pm or Wednesday morning would work well."

**Demo Hand, engineer:** "quick sync this week? tue after 2pm or wed morning"

**AI draft:** "Would you be available for a quick sync sometime this week? I'm flexible and happy to work around your schedule. Tuesday after 2pm or Wednesday morning would work well."

**Demo Hand, designer:** "Quick sync this week? Tuesday after 2pm or Wednesday morning works for me."

Same meeting, same two slots. The engineer writes "tue", skips the capitals and stops at the times. The designer spells out the days and adds "works for me". A style summary would likely call both of them "casual and brief" and give each the same message.

## Running both

1. Keep your skill on in Customize > Skills for long-form work in Claude.
2. Add Ownhand as a custom connector under Customize > Connectors, or in Claude Code run `claude mcp add --transport http --scope user ownhand https://ownhand.dev/mcp`.
3. Build your Hand from things you actually sent. AI drafts don't count as samples.
4. Paste the agent blurb from the [docs](https://ownhand.dev/docs#blurb) into Instructions for Claude, so Claude knows to call Ownhand for outgoing messages and to report what you sent.

If you are weighing a plain style note against Ownhand, [Ownhand vs custom instructions](https://ownhand.dev/answers/ownhand-vs-custom-instructions) covers that side.

## Follow-up questions

### What happened to the custom style I made in Claude?

Anthropic's notice said custom styles migrate to skills automatically and arrive switched off. Open Customize > Skills and turn yours on. In a chat, call it with a slash command made from its name: lowercase, spaces turned into hyphens, punctuation dropped, and -style on the end. A style called Board Updates becomes /board-updates-style.

### Can I use a Claude skill and Ownhand at the same time?

Yes. A skill shapes how Claude answers you in the chat. Ownhand only rewrites text you are about to send as yourself, when your agent calls its write tool. Keep the skill for long documents you draft with Claude, and let Ownhand handle the Slack message or email that leaves the chat.

### Does Ownhand work in claude.ai, or only in Claude Code?

Both. In Claude on the web or the desktop app, add it as a custom connector under Customize > Connectors. Anthropic allows one custom connector on the free plan. In Claude Code it is one command, shown at the bottom of this page.

### Does Ownhand read my Claude chats?

No. It only sees what your agent sends with a write call: the draft, the occasion, and for a reply, up to 30 messages from the thread, which are used for that request and never stored. Your samples, the style card built from them and your learned rules are kept with your account.

### Which one costs less?

A skill costs nothing beyond your Claude plan, and skills are on Free, Pro, Max, Team and Enterprise as long as code execution is on. Ownhand gives new accounts free credit and then charges per rewrite and per feedback call, with no subscription.

## Sources

Checked on 4 October 2026. If something here is out of date, email [hi@thalientlabs.ai](mailto:hi@thalientlabs.ai) and we will fix it.

- [Claude Help Center: Styles are moving to skills (Wayback copy of 10 July 2026)](https://web.archive.org/web/20260710101210/https://support.claude.com/en/articles/10181068-styles-are-moving-to-skills): last updated 1 June 2026; custom styles become skills, off by default; slash command naming; Concise, Explanatory and Formal removed; Learning kept as a skill; styles menu removed after the migration. The page no longer loads on the live help center.
- [Claude Help Center: Configuring and using styles (Wayback copy of 15 January 2026)](https://web.archive.org/web/20260115223141/https://support.claude.com/en/articles/10181068-configuring-and-using-styles): how styles worked: presets, a custom style from uploaded or pasted writing samples (PDF, DOC, TXT), or from a description, with instructions you could edit by hand
- [Claude Help Center: Claude's personalization features](https://support.claude.com/en/articles/10185728-understanding-claude-s-personalization-features): updated 16 September 2026; lists Instructions for Claude, project instructions and skills, with no styles
- [Claude Help Center: What are skills?](https://support.claude.com/en/articles/12512176-what-are-skills): plans, code execution requirement, skills vs projects and custom instructions, Agent Skills open standard
- [Claude Help Center: Use skills in Claude](https://support.claude.com/en/articles/12512180-use-skills-in-claude): Customize > Skills, toggling skills, uploading a skill as a ZIP
- [Claude Help Center: How to create custom skills](https://support.claude.com/en/articles/12512198-how-to-create-custom-skills): skill.md with a name and a description of up to 200 characters, reference files, include examples
- [Claude Help Center: custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp): Customize > Connectors; free users get one custom connector
- [Ownhand source code](https://github.com/jordan-gibbs/ownhand): occasions, the reply fingerprint, the checker and the feedback loop

## Install

In Claude Code it is one line. You sign in through your browser, so there is no API key to copy.

```
claude mcp add --transport http --scope user ownhand https://ownhand.dev/mcp
```

For Claude, ChatGPT, Cursor, VS Code, Codex, Gemini CLI, Windsurf, Zed, Goose and Cline, see [Connect](https://ownhand.dev/connect). New accounts get $1 of free credit. After that you pay as you go, and each rewrite costs about half a cent.

More answers: [all questions](https://ownhand.dev/answers). Full reference: [Docs](https://ownhand.dev/docs).

---

Page: https://ownhand.dev/answers/ownhand-vs-claude-styles. Index for agents: https://ownhand.dev/llms.txt