MCP server

Talk to Post Mate from Claude, Cursor, ChatGPT, or any MCP-compatible client. Schedule posts in plain English.

The Model Context Protocol (MCP) is the open standard Anthropic introduced for connecting AI clients to external tools. Post Mate ships an MCP server out of the box — point your client at it, authenticate once, and you can ask the model to schedule posts, list connections, read analytics, all in natural language.

What it looks like in practice

In Claude Desktop:

"Schedule a post to my Instagram, TikTok, and LinkedIn tomorrow at 10am: 'The grid project is live → post-mate.com/tools/instagram-grid-maker'"

Claude calls our create_post MCP tool with the right arguments, we schedule the post, and the model replies with a confirmation and the post ID.

Server URL

https://post-mate.com/api/mcp

It's a standard MCP HTTP server. Compatible clients today: Claude Desktop, Cursor IDE, ChatGPT (in MCP mode), Continue, and a growing list of others.

Authentication

The MCP server uses Bearer authentication, same key as the REST API. Generate one at /app/settings/api — same flow, no separate "MCP key" concept.

The key needs posts:write scope to schedule, posts:read to list, accounts:read to enumerate connections.

Claude Desktop setup

Edit your Claude Desktop config (location varies by OS):

{
  "mcpServers": {
    "post-mate": {
      "url": "https://post-mate.com/api/mcp",
      "headers": {
        "Authorization": "Bearer pm_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

Restart Claude Desktop. The Post Mate tools should show up in the slash-command picker. Test with: "list my connected social accounts."

Cursor IDE setup

Cursor → Settings → MCP → Add MCP server:

Name: post-mate
URL: https://post-mate.com/api/mcp
Header: Authorization: Bearer pm_live_…

After Cursor reconnects, the Post Mate tools are usable in chat and inline AI prompts.

ChatGPT setup

ChatGPT's MCP support is in the Custom GPT builder. Add a new action, paste the server URL, paste your Bearer key. The action catalog populates from our MCP descriptor automatically.

Tools exposed

The MCP server exposes the same surface as the REST API, but framed as discrete tools the model can call:

Tool	What it does
`create_post`	Schedule a new post
`list_posts`	List posts with filters
`get_post`	Read one post by ID
`update_post`	Edit a scheduled post
`cancel_post`	Cancel a scheduled post
`list_accounts`	Enumerate connected social accounts
`get_analytics`	Read post-level metrics
`list_queue`	Read the queue schedule

Each tool ships with a JSON schema for arguments so the model knows exactly what to pass without guessing.

Cost

The MCP server is included in the same $5/mo API add-on as the REST API and webhooks. No separate billing tier.

Why we built this

Most cross-posting tools require you to open a dashboard, navigate to compose, click through fields. MCP lets you write a sentence instead. We use it ourselves for ~30% of our own scheduling volume; it's genuinely faster for one-shot posts and trivially faster for analytics queries.

On this page