LinkedIn MCP server: draft and schedule posts from Claude

LinkedIn MCP server: what the assistant does

MCP’s documentation describes tools as executable functions an AI application can invoke to perform actions. Dripost’s tools for LinkedIn are the same ten used everywhere else: see your channels, list and read posts, list media, create, update and delete a post, import media from a link, publish, and retry.

For LinkedIn that means a thought-leadership draft written to your brief, edited after your notes, and queued for the morning you choose.

1. Connect your LinkedIn profile

Add LinkedIn under Channels in Dripost. The target is your personal profile; the assistant can see it, as a connected channel, when it asks for the workspace.

2. Connect Dripost to your assistant

Claude takes a remote server as a custom connector: add mcp.dripost.com/mcp under Customize, Connectors. Claude’s help says it asks for approval before invoking a tool, and you can allow each action or choose Allow always for a server you trust.

For ChatGPT, use Developer mode and add the same address. Both Claude and ChatGPT have been connected to Dripost and tested end to end.

3. Ask, review, release

Try “Draft three LinkedIn posts from this article link and queue the best one for Tuesday.” Review the draft in Dripost, where you can reword anything before it is sent.

Scheduling needs the separate “post” permission, which you tick; without it the assistant can only make drafts.

The holds that apply

A post the AI wrote or changed waits 10 minutes before it goes out, so you can cancel it. Times an app sets must be at least 10 minutes ahead, and “publish now” needs your approval on dripost.com.

When the post goes out you get the link to it. If LinkedIn refuses it, the refusal arrives with LinkedIn’s own reason, and retrying touches only that platform.

Sources

• Model Context Protocol: architecture overview
• Claude Help Center: custom connectors using remote MCP