Publish to LinkedIn from an AI agent with MCP

Inklate ships a built-in Model Context Protocol (MCP) server, so any MCP-capable client — Claude Code, Claude Desktop, Cursor, or an agent you build yourself — can operate your content pipeline. Here’s how to go from zero to a published LinkedIn post.

1. Mint an organization API key

In Inklate, open your organization settings and create an organization API key. The key is prefixed inklate_ and scopes every call to that one organization, so your agent never has to pass an organization argument.

2. Point your MCP client at Inklate

Add Inklate to your client’s MCP configuration:

{
  "mcpServers": {
    "inklate": {
      "url": "https://api.inklate.com/mcp",
      "headers": { "Authorization": "Bearer inklate_your_org_key" }
    }
  }
}

That’s the same guarded surface the dashboard uses. Membership is re-checked on every call, so a revoked key fails closed.

3. Connect a channel

Channels are publishable destinations — a LinkedIn profile or company page. Connect one in the dashboard once; after that your agent can list channels and publish to them.

4. Let the agent publish

Now you can ask your agent to do the work:

Draft a post announcing our launch, schedule it for 9am tomorrow in my timezone, and publish it to our company page.

The agent drafts the post, schedules it (Inklate requires an explicit UTC offset so the time can’t be misread), and the durable publish worker sends it at the right moment. Later, ask the agent how the post performed and it reads back the latest metrics.

Why this matters

Because the dashboard and the agent share one API, there’s no second-class automation path to maintain. Whatever your team can do by hand, your agents can do at scale — and everything stays inside the same multi-tenant, permissioned organization.

Ready to try it? Create a free account.