pipeboard-co/meta-ads-mcp

What it solves

This project provides a way for AI assistants (like Claude, ChatGPT, and Cursor) to manage Meta Ads campaigns end-to-end through natural language. It eliminates the need for manual dashboard navigation by allowing users to launch campaigns, upload creatives, and analyze performance across Facebook and Instagram via a conversational interface.

How it works

It is a Model Context Protocol (MCP) server that connects AI clients to Meta's public APIs. It can be deployed as a hosted remote MCP for immediate use without developer tokens, or installed locally for technical users. The server exposes a set of tools (such as create_campaign and get_campaigns) that the AI agent uses to execute actions in the Meta Ads Manager.

Who it’s for

• Marketers who want to manage and optimize their ad spend using AI assistants.
• AI Power Users using MCP-compatible clients like Claude Desktop or Cursor to automate advertising workflows.
• Developers who want to integrate Meta Ads management capabilities into their own AI agents.

Highlights

• End-to-End Management: Full control over campaigns, ad sets, and ads, including budget updates and targeting changes.
• Creative Operations: Ability to upload images and build creatives (including A/B testing) directly through the AI client.
• Cross-Platform Ecosystem: Part of a larger family of MCP servers covering Google, TikTok, Snap, and Reddit Ads under a single authentication flow.
• Safety Guardrails: Implements an explicit confirmation model for all write actions and defaults new campaigns to a paused state.
• Partner Backed: Developed by a badged Meta Business Partner.