Google has open-sourced a Model Context Protocol (MCP) server that exposes read-only access to the Google Ads API for agentic and LLM applications. The repository googleads/google-ads-mcp implements an MCP server in Python that surfaces two tools today: search (GAQL queries over Ads accounts) and list_accessible_customers (enumeration of customer resources). It includes setup via pipx, Google Ads developer tokens, OAuth2 scopes (https://www.googleapis.com/auth/adwords), and Gemini CLI / Code Assist integration through a standard MCP client configuration. The project is labeled “Experimental.”
So, why it matters?
MCP is emerging as a common interface for wiring models to external systems. By shipping a reference server for the Ads API, Google lowers the integration cost for LLM agents that need campaign telemetry, budget pacing, and performance diagnostics without bespoke SDK glue.
How it works? (developer view)
- Protocol: MCP standardizes “tools” that models can invoke with typed parameters and responses. The Ads MCP server advertises tools mapped to Google Ads API operations; MCP clients (Gemini CLI/Code Assist, others) discover and call them during a session.
- Auth & scopes: You enable the Google Ads API in a Cloud project, obtain a developer token, and configure Application Default Credentials or the Ads Python client. Required scope is
adwords. For manager-account hierarchies, set a login customer ID. - Client wiring: Add a
~/.gemini/settings.jsonentry pointing to the MCP server invocation (pipx run git+https://github.com/googleads/google-ads-mcp.git google-ads-mcp) and pass credentials via env vars. Then query via/mcpin Gemini or by prompting for campaigns, performance, etc.
Ecosystem signal
Google’s server arrives amid broader MCP adoption across vendors and open-source clients, reinforcing MCP as a pragmatic path to agent-to-SaaS interoperability. For PPC and growth teams experimenting with agentic workflows, the reference server is a low-friction way to validate LLM-assisted QA, anomaly triage, and weekly reporting without granting write privileges.
Key Takeaways
- Google open-sourced a read-only Google Ads API MCP server, showcasing two tools:
search(GAQL) andlist_accessible_customers. - Implementation details: Python project on GitHub (
googleads/google-ads-mcp), Apache-2.0 license, marked Experimental; install/run viapipxand configure OAuth2 with thehttps://www.googleapis.com/auth/adwordsscope (dev token + optional login-customer ID). - Works with MCP-compatible clients (e.g., Gemini CLI / Code Assist) so agents can issue GAQL queries and analyze Ads accounts through natural-language prompts.
Conclusion
In practical terms, Google’s open-sourced Google Ads API MCP server gives teams a standards-based, read-only path for LLM agents to run GAQL queries against Ads accounts without bespoke SDK wiring. The Apache-licensed repo is marked experimental, exposes search and list_accessible_customers, and integrates with MCP clients like Gemini CLI/Code Assist; production use should account for OAuth scope (adwords), developer token management, and the data-exposure caveat noted in the README.
Check out the GitHub Page and technical blog. Feel free to check out our GitHub Page for Tutorials, Codes and Notebooks. Also, feel free to follow us on Twitter and don’t forget to join our 100k+ ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Michal Sutter is a data science professional with a Master of Science in Data Science from the University of Padova. With a solid foundation in statistical analysis, machine learning, and data engineering, Michal excels at transforming complex datasets into actionable insights.
Credit: Source link
