What Is the Meta Ads MCP Server?

MCP stands for Model Context Protocol. Meta runs its own MCP server for ads, so an AI application can connect directly to Meta’s advertising capabilities and work through natural language instead of custom integration code. Meta’s developer documentation sums it up in one line: the Ads MCP Server lets you manage Meta ads from an AI agent.

What can an agent do through it?

Meta’s announcement lists these use cases and says it keeps adding tools:

Area	What Meta says the agent can do
Ads creation and editing	Create, edit and delete campaigns, ad sets and ads
Custom audiences	Create, update and delete custom audiences
Catalogs	Create a catalog, add products, set up product sets and fix issues stopping products from showing in ads
Performance	Query spend, impressions, CTR, ROAS and more, with date ranges, age, gender and platform breakdowns, and several aggregation levels
Signal diagnostics	See signal health and quality information
Help and troubleshooting	Search Meta Business Help Center articles
Tests	Create and manage A/B tests and conversion lift studies, and read existing ones

You can limit what an agent does. Meta’s ads MCP server rules define the actions an agent can and cannot take on an ad account or catalog, for example read access to performance data without the ability to change budgets, campaigns or creative.

Who can use it?

On 16 July 2026 Meta opened the server to any developer with their own Meta app, so they can bring it into tools their customers already use, such as a marketing platform, a data warehouse or an internal system. Meta also says it’s gradually rolling out access to the tools available to ad accounts through this method, and suggests asking your agent which tools you have.

How does a developer connect an app?

Meta’s steps from its announcement:

1. Go to the developer dashboard and select your app.
2. Click Use cases in the left navigation, then Add use cases.
3. Click Ads and monetization, select Create & manage ads with ads MCP server, and click Save.
4. Choose how to authenticate: OAuth through Facebook Login for Business, or a pre-obtained access token.

Agencies and partners that manage data for other businesses need one more step: submit the app for review and request Advanced Access on the ads_mcp_management permission.

What does the ads_mcp_management permission cover?

Meta’s permissions reference says it lets an app access the Meta ads MCP server so AI agents can interact with Meta Ads on behalf of advertisers. Allowed uses are creating, editing and managing campaigns, retrieving performance insights and reports, and managing business assets such as catalogs, ad accounts and pixels.

FAQ

Do I need to write code to use it?

Meta says that before the server, connecting AI agents to Meta ads meant writing and maintaining custom integration code, and that developers can now skip that. You still need a Meta app set up with the ads MCP use case, and an authentication method, either OAuth or an access token.

What does App Review ask for?

For the ads_mcp_management permission, Meta asks for specific examples of why your app needs the server to manage campaigns and business assets for other businesses, plus a screencast. It must show the full Facebook login flow granting the permission, a business reaching its ads performance data in your app, and data such as impressions, conversions, spend, clicks and reach displaying.

Is it the same as Meta Social Technologies MCP?

No. Meta Social Technologies MCP, formerly Meta Developer Tools MCP, is a separate server for developers managing the Meta apps they build: reading app configuration, monitoring API health, checking App Review status, configuring webhooks and searching developer documentation. The ads MCP server is the one for managing ads.