Advertising MCP stack for your AI
Three ad networks, the analytics property they all claim credit in, and the store that settles it.
- Meta Ads
- Google Ads
- TikTok Ads
- Google Analytics 4
- Shopify
app.popmcp.com/mcpThe 5 connections
Every ad platform reports on itself, in its own attribution window, and the three of them together always add up to more than the money that actually arrived.
- Meta AdsAPI credentialCatalogsAccountAudiencesBusiness224tools22triggers
- Google AdsOAuthAccountAdsCampaignsKeywords21tools14triggers
- TikTok AdsOAuthBusiness centerCreativeCatalogMeasurement431tools19triggers
- Google Analytics 4OAuthAdminReporting21tools7triggers
- ShopifyOAuthOrdersCollectionsCustomersInventory70tools24triggers
Across the stack437 read and 330 write operations, 79 API modules767tools86triggers
One question, every connection
The tool names below are the real ones your client calls. Check any chain against the catalog on the integration’s own page.
Which network actually earned the last $10k of spend?
- Meta Ads
meta_ads_get_campaign_insights - Google Ads
google_ads_search_stream - TikTok Ads
tiktok_ads_get_report - Google Analytics 4
ga4_run_report - Shopify
shopify_get_orders
Three platform reports set against one analytics property and the real order total, so the double-counting is visible instead of averaged away.
Pause everything over a $40 cost per result and tell me what is left running.
- Meta Ads
meta_ads_list_adsetsmeta_ads_get_adset_insightsmeta_ads_update_adset - Google Ads
google_ads_searchgoogle_ads_mutate_campaigns - TikTok Ads
tiktok_ads_get_campaignstiktok_ads_update_campaign_status
One threshold applied across three ad managers in one instruction, then read back so you can see what survived.
Which products are worth scaling into this week?
- Shopify
shopify_get_productsshopify_get_orders - Google Analytics 4
ga4_batch_run_reports - Meta Ads
meta_ads_list_campaignsmeta_ads_get_campaign_insights
The scaling decision starts from what sold and what margin is left, not from whichever creative the platform is proud of.
Turning it on
Nothing to deploy, and nothing to install on the provider side.
- Connect the accounts
One connection per store, property or ad account. Four authorise with OAuth, one take an API credential. Credentials are encrypted at rest and never reach your AI client.
- Tick what each client reaches
Your client signs in to PopMCP with OAuth and gets only the connections you approve. Writes run the moment it calls them, so keep reporting teammates on read-only.
- Register one URL
app.popmcp.com/mcp, once per client. Claude, ChatGPT, Codex and the rest reach the same five connections with the same permissions.
Five connections need Starter or above — the free tier holds two. See what each plan includes.
The other stacks
Same endpoint, same controls. Most teams end up running two or three of them side by side.
Questions about the advertising stack
Does this replace my ads manager?
No. It removes the tab-switching from the questions that cross networks. Building a campaign from scratch is still faster in the native UI; comparing three networks, finding what is burning budget and pushing the same change to all of them is what this is for.
Can one connection cover several ad accounts?
Google Ads reads across the MCC tree from one connection. Meta Ads and TikTok Ads are connected per ad account, so an agency running twelve clients has twelve connections — which is the point of the workspace and connection limits on each plan.
What stops an agent from changing a budget it should not?
Two things, both yours to set. A client only reaches the connections you tick on the PopMCP consent screen, and on Scale a teammate can be held to read-only per connection, which hides every write tool including the budget and status mutations.
Connect the advertising stack to your AI
Add Meta Ads, Google Ads, TikTok Ads and the rest with credentials you already own, and get one secure MCP URL for all of them.





