Ampost MCP Server: Social Media for AI Agents

AI agents should be able to post to social media just like humans do. Today we're releasing the Ampost MCP Server — a Model Context Protocol implementation that lets LLMs and AI agents create, schedule, and manage posts across every supported platform.

Why MCP matters

Model Context Protocol (MCP) is an open standard that lets AI systems discover and use external tools through a structured interface. Instead of building custom integrations for every AI framework or hardcoding API calls into prompt engineering, you point your agent at our MCP server and it gains the ability to manage social content natively.

This means Claude Desktop, Cursor, or any MCP-compatible agent can now publish posts, check status, and manage platform connections without you writing a single line of glue code. If you want the concrete setup details, the full MCP server documentation walks through authentication, available tools, and payload examples.

Available tools

The Ampost MCP server exposes a complete toolkit for social media automation:

• platforms_list — Read each platform's rules: text limits, media requirements, placements and options.
• posts_create — Create a post with text, media, and per-platform options. Supports drafts, immediate publishing (publishNow) and scheduling (scheduledFor).
• posts_publish and posts_retry — Publish a draft now, or retry the accounts that failed.
• posts_list and posts_get — Check posts and each platform's result.
• posts_update and posts_delete — Change or discard a post before it goes live.
• accounts_list and accounts_get — View connected accounts and their token health.
• accounts_connect — Start a connection; the user finishes consent in their browser.
• platform_sets_list and friends — Browse and manage platform sets, the groups of accounts you publish to.

Every tool is one endpoint of the REST API, with the same input, output and error codes. The MCP docs list them all.

Each tool returns structured data that the LLM can reason about, so agents can make decisions like "Instagram failed — should I retry or alert the user?"

How to connect

The easiest way is through your agent's MCP configuration. Add Ampost to Cursor or any client that supports HTTP servers with headers; Connect Ampost to your AI agent has the command or config for each client, including Claude Code and the Claude app:

{
  "mcpServers": {
    "ampost": {
      "url": "https://api.ampost.io/mcp",
      "headers": {
        "Authorization": "Bearer amp_live_..."
      }
    }
  }
}

Once configured, your agent can ask you things like:

"I see you have Instagram and Facebook connected. Should I post this update to both platforms?"

Security and control

The MCP server uses the same API key authentication as the REST API documentation, or an OAuth sign-in limited to the scopes and platform sets the user allows. It respects your plan and rate limits. Agents never see social platform credentials: accounts_connect returns a link, and the user grants access in their own browser. Billing and API keys stay in the dashboard.

Use cases

• Content calendars that auto-generate and schedule posts based on your editorial pipeline
• Social media managers that respond to trends in real time by monitoring RSS feeds or APIs
• Brand monitoring agents that publish updates when product releases, blog posts, or milestones occur
• Customer support agents that cross-post announcements or status updates across all channels

What's next

Since launch the toolkit has grown to analytics, media uploads, posting queues, CSV import and webhooks; the MCP docs list every tool. The goal is simple: any AI agent that understands language should be able to manage social media effectively. For the broader product surface around scheduling and multi-platform publishing, see the features overview or the getting started guide.

Install the server today and let your agents start posting.