GuidesMeta Ads MCP

Meta Ads MCP Server

Connect Claude, ChatGPT, or any AI tool to your Meta Ads account. Manage campaigns, optimize creatives, and analyze performance with natural language.

What is the Meta Ads MCP?

The Pipeboard Meta Ads MCP (Model Context Protocol) Server is a source-available server that connects AI assistants like Claude, ChatGPT, and custom agents to the Meta Ads API. Manage campaigns, optimize ad creatives, analyze performance, and control targeting through natural language.

Instead of building API integrations from scratch, you get instant access to campaign management, analytics, targeting tools, creative upload, and more.

Key features:

30+ pre-built tools covering the full Meta Ads management lifecycle
Works with Claude Desktop, ChatGPT, Python, n8n, and any MCP client
Creative management with image upload, carousel ads, and A/B testing
Audience targeting with interests, behaviors, demographics, and geo-locations
Manage at scale: bulk-update budgets and status across many ads, and one-click duplicate campaigns, ad sets, ads, and creatives
Safety-first: campaigns created in PAUSED status by default
Different from Facebook’s MCP at mcp.facebook.com/ads. That one is in a gated beta and returns is_ads_mcp_enabled: false for most ad accounts. Pipeboard’s MCP uses the standard Meta Marketing API and works on every account you have access to. More on the Facebook MCP rollout →

Getting Started

Set up in under 2 minutes

1. Sign up for a Pipeboard account:

pipeboard.co/auth/signup →

2. Connect your Meta Ads account:

pipeboard.co/connections →

3. Add the MCP server to Claude (or another MCP client):

https://meta-ads.mcp.pipeboard.co/

Use this URL as a Streamable HTTP MCP server in your client. Authentication is handled automatically through your Pipeboard session.

Connect to Your AI Tool

The Meta Ads MCP works with any MCP-compatible client. Just add the server URL and start managing your campaigns with natural language.

Claude Pro / Max

Recommended
  1. 1.Open claude.ai/customize/connectors
  2. 2.Click "Add custom connector"
  3. 3.URL: https://meta-ads.mcp.pipeboard.co/

ChatGPT

Developer mode
  1. 1.Enable Developer mode in ChatGPT Settings
  2. 2.Settings → Apps → Advanced → Developer mode
  3. 3.Create app with MCP URL

Claude Code

CLI / VS Code

Run this command in your terminal:

claude mcp add --transport http pipeboard-meta-ads https://meta-ads.mcp.pipeboard.co/

Then type /mcp in Claude Code to authenticate with Pipeboard.

Alternative: use an API token for instant setup (no browser auth needed):

claude mcp add --transport http pipeboard-meta-ads https://meta-ads.mcp.pipeboard.co/?token=pk_...

Get your token at pipeboard.co/api-tokens

Claude Desktop JSON configuration:

Add this to your Claude Desktop MCP settings (Settings → MCP Servers → Add):

claude_desktop_config.json
{
  "mcpServers": {
    "meta-ads": {
      "url": "https://meta-ads.mcp.pipeboard.co/"
    }
  }
}

Tips for getting started

Browse our prompt library for ready-to-use prompts you can copy into Claude
Campaigns are created in PAUSED status by default so you can review before going live

Try these prompts

Campaign Overview

"List my Meta ad accounts and show campaign performance for the last 30 days"

Audience Research

"Search for interest targeting options related to fitness and estimate the audience size"

Creative Analysis

"Show me the creatives for my top-performing ads and suggest improvements"

Budget Optimization

"Analyze spend vs conversions across campaigns. How should I redistribute budget for better ROAS?"

Open Source

The MCP server is source-available on GitHub

Review the code, contribute features, or run your own instance. Built with Python and the official MCP SDK.

Best Practices

  • Create campaigns in PAUSED status, then review before activating
  • Batch operations when possible — use filters to reduce API calls
  • Test targeting with audience estimation before launching campaigns
  • Use breakdowns (age, gender, platform) to find top-performing segments
  • Cache insights data for reporting workflows
  • Use the Meta API rate limit of 200 calls/hour wisely — the MCP handles retries automatically

Pipeboard vs Meta’s official MCP

Both connect AI to Meta Ads. Here’s where they differ.

Pipeboard Meta Ads MCPMeta official connector
Ad platformsMeta, Google, TikTok, Snap & RedditMeta only
Account availabilityEvery account you can access (standard Marketing API)Gated beta — is_ads_mcp_enabled: false for most accounts
SafetyCampaigns paused by default — you review before anything goes liveWrites directly to your live ad account
AI clientsClaude, ChatGPT, Cursor, Claude Code, n8n, any MCP clientMCP-compatible clients
Source-availableYes — auditable on GitHubNo
PriceFree planFree

Pipeboard creates every campaign paused by default, so an AI assistant can draft and propose changes while you keep final approval.

See the full field compared in Best Meta Ads MCP servers (2026) — Pipeboard vs Meta’s official MCP, Markifact, Adspirer, and Ryze — or read why Pipeboard is the most reliable ads MCP (reads + writes, reliable duplication, bulk edits across 8 platforms).

Frequently asked questions

What is the Meta Ads MCP?

The Pipeboard Meta Ads MCP is a source-available Model Context Protocol server that connects AI assistants — Claude, ChatGPT, Cursor, or any MCP client — to your Meta (Facebook and Instagram) Ads account. It exposes 30+ tools so you can launch campaigns, upload creatives, adjust budgets, and analyze performance in natural language.

How is Pipeboard different from Meta's official Ads MCP?

Meta's official connector (mcp.facebook.com/ads) is in a gated beta and returns is_ads_mcp_enabled: false for most ad accounts, and it only covers Meta. Pipeboard uses the standard Meta Marketing API so it works on every account you can access, creates campaigns paused by default for your review, is source-available, and is part of one MCP family that also covers Google, TikTok, Snap, and Reddit Ads.

Is it safe to let an AI manage my ad account?

Pipeboard creates campaigns in PAUSED status by default, so nothing goes live until you review and activate it. You connect your own ad account and stay in control of access, and the server is source-available so you can audit exactly what it does.

Does the Meta Ads MCP work with Claude and ChatGPT?

Yes. It works with Claude (Pro/Max and Claude Code), ChatGPT (developer mode), Cursor, n8n, Python, and any MCP-compatible client. You add one Streamable HTTP URL (https://meta-ads.mcp.pipeboard.co/) and authenticate through your Pipeboard session.

Is the Pipeboard Meta Ads MCP free?

Yes, there is a free plan. You can connect your Meta Ads account and start managing campaigns at no cost; paid plans add higher limits and more connected accounts. See pipeboard.co/pricing.

Can I manage other ad platforms too?

Yes. Pipeboard is one MCP family that covers Meta, Google, TikTok, Snap, and Reddit Ads, so you can manage campaigns across all of them from the same AI assistant. (The Meta Ads MCP itself is source-available on GitHub.)

Can it bulk-edit or duplicate campaigns?

Yes. The Meta Ads MCP includes bulk operations and one-click duplication that the Meta UI makes tedious. bulk_update_ads and bulk_update_adsets change budgets, bids, or status across many objects at once, and bulk_get_insights pulls performance for many ads in a single call. duplicate_campaign, duplicate_adset, duplicate_ad, and duplicate_creative clone a winning setup — optionally with copy variations — so you can scale it across audiences. Duplicated and bulk-created campaigns follow the same paused-by-default safety rule.

How long does setup take?

Under two minutes: sign up for Pipeboard, connect your Meta Ads account, and add the server URL https://meta-ads.mcp.pipeboard.co/ to Claude or your MCP client.

Ready to connect your AI tools to Meta Ads?

Get started with the Meta Ads MCP in under 2 minutes.