Claude drafts posts in the voice of your recent ones, fits them to X's limit, and schedules or publishes them to your account once you approve, then, on Pro or during its 7-day trial, shows you how they did.
Free for 2 social accounts. No card, no API key.
Free plan works. It allows one custom connector.
On a phone? In the Claude app: Settings, Connectors, Add custom connector. Paste the same name and URL.
Free plan works. It allows one custom connector.
Ask Claude things like these. It shows you the post before anything is scheduled or published.
“Announce the autumn menu on X with these two photos. Keep it to one line.”


Attach up to four images, a GIF or a video. Claude keeps the text inside the 280 characters X allows.
“Cut it to one line.”
Claude reads your last 15 posts first, so it keeps your length, tone and hashtag habits.
“Schedule it for 9am tomorrow.”
ScheduledTomorrow · 09:00Publish now or give a day and time. It sits in Antwork's calendar, where you can move or cancel it.
“Which of my posts this month got the most replies?”
Antwork reads each post's numbers back from X after it goes out.
“My post failed to publish. Find out why and retry it.”
A failed post comes back with the reason. Retrying it is one message.
Yes, through a connector.
Claude can post to X (formerly Twitter) through Antwork, a hosted Model Context Protocol (MCP) connector. You add Antwork's server URL as a custom connector in Claude, sign in to Antwork, and approve X once on X's own OAuth screen. From then on Claude reads your recent posts on X, drafts new ones in that voice, counts them the way X does, and schedules or publishes them to your account from inside the conversation, through X's official API. There is no X developer account to apply for, no API key to paste and no browser extension driving your logged-in x.com.
X MCP servers compared: which post with their own approved app, and which need your developer keysClaude checks these before it drafts, so a post does not fail at publish time. They come from the same rules Antwork's publisher enforces.
Each works. They differ in where the work happens and how much you set up.
| Antwork connector | An automation tool | Build your own | |
|---|---|---|---|
| Where you ask | In the Claude chat | In a workflow you build | In your own code |
| Setup | Paste one URL, sign in | Design a scenario per task | Developer app, API access, hosting |
| Writes in your voice | Reads your recent posts | Only what you wire in | Only what you build |
| Schedules and publishes | Yes, from the chat | On the workflow's trigger | If you build it |
| Shows the post first | Every time | Depends on the workflow | Depends on your build |
| Code needed | None | None | Yes |
An automation tool is the better fit when a post should be triggered by something outside the chat, like a new row in a spreadsheet. If you want to ask Claude and see the post, the connector is the shortest path.