Google Ads MCP server for your AI

Your own Google OAuth client and a developer token. The connect form asks for an OAuth Client ID, an OAuth Client Secret and a Developer Token, then hands you to Google to authorize once. After the authorization you choose which accounts to connect and which workspace each one belongs in.

No. Manager (MCC) hierarchies are resolved during the connect flow, so there is no Customer ID to go and find. You pick the accounts you want from what the authorized login can reach, and list_accessible_customers shows your agent that same set later.

Both, and you choose which. With full access it can pause or edit campaigns through mutate_campaigns, move money with mutate_campaign_budgets, manage keywords and negatives with mutate_ad_group_criteria, and act on Google's suggestions with apply_recommendation or dismiss_recommendation. Set the connection to read-only and all of that goes away, leaving reporting; keep writes on with approval required and each change waits for your sign-off.

With GAQL. search runs a query against the account and pages through the results, while search_stream returns the entire result set in one response. When the agent does not know the field names, search_fields discovers queryable fields, resources, metrics and segments, and get_field returns metadata for a single field by name. Search-volume work takes a different route, through generate_keyword_ideas and generate_keyword_historical_metrics.

mutate_ads maps to AdService.MutateAds, which is update-only — it changes the creative content of ads that already exist. New ads are created at the ad group level through mutate_ad_group_ads. When several changes need to land together, mutate is the universal batch call: create, update and remove across any services in one atomic call.

21 operations across 12 API modules, 7 read and 14 write, and all 21 load by default rather than hiding behind a search. The largest groups are Account, Recommendations, Planning, Metadata, Ads, Keywords, Campaigns and Reporting. There are also 14 triggers, split between change detection (campaign created, campaign status changed, ad disapproved, new search term) and thresholds (daily spend spiked, cost per conversion spiked, campaign stopped serving, impression share lost to budget).

No. PopMCP hosts it and gives you a Streamable HTTP MCP URL. There is nothing to deploy, no callback URL to register, and no infrastructure to keep running or patch.

Any client that supports a remote Streamable HTTP MCP server — Claude Desktop, Claude Web, ChatGPT custom connectors, Codex and the Codex CLI, OpenCode and Z.ai's ZCode, as well as editors such as Cursor, Windsurf and Zed. Your client signs in to PopMCP with OAuth; the provider credential never leaves the server.

Loading 21 tool definitions into every conversation burns context before the work starts. PopMCP loads a curated 21-tool set covering the everyday workflows and keeps the rest searchable, so your AI pulls a specialist tool only when a task needs it. You can switch a connection to the full catalog at any time.

A tool is something your AI calls — there are 21 here. A trigger is something PopMCP watches for you: 14 Google Ads events that can start work on the PopMCP side. Triggers never enter your AI client's context, and none of them ask you to paste a callback URL into Google Ads.