google_ads_mcp

The Google Ads MCP Server is a Model Context Protocol (MCP) service that exposes the Google Ads API to large language models (LLMs) via an MCP interface. This enables LLMs to issue structured requests to manage campaigns, ad groups, and other Google Ads resources through a standardized MCP endpoint. To use it, deploy the server locally or in your environment, then integrate it into your Gemini (or other MCP clients) configuration. The README shows how to configure Gemini to discover the Google Ads MCP server and how to call commands like listing campaigns or fetching metrics. Typical usage involves running the server with uv and pointing your MCP client to the server, optionally providing the path to your Google Ads credentials via an environment variable. Ensure your google-ads.yaml credentials file is accessible to the server (by default in your home directory, or via an explicit path you provide). Once running, you can query capabilities such as listing campaigns, retrieving performance metrics, or fetching ad group data through natural language prompts mapped to Google Ads API calls by the MCP server.

Prerequisites

• Python 3.12 installed on your system
• Either pipx or the uv tool available for dependency and environment management

Installation steps

1. Install Python prerequisites

• macOS/Linux: ensure Python 3.12 is installed. On many systems you can use pyenv or your system package manager.

2. Install uv (recommended for this project)

• If you have uv instructions available, install uv and then install dependencies via the project workflow. Example (via uv):

# If you already have uv installed, install or sync dependencies for this project
uv pip sync

• If you prefer pipx, you can install uv via your preferred method and then run similar commands to install dependencies.

3. Prepare Google Ads credentials

• Ensure you have a google-ads.yaml file with the following keys: client_id, client_secret, refresh_token, developer_token, and optionally login_customer_id. Place the file where the server expects it (default: your home directory) or provide an explicit path via an environment variable if supported by your setup.

4. Run the MCP server

• Direct use (via uv):

uv run -m ads_mcp.server

• The server will start and be ready to accept MCP requests. Integrate with Gemini or your MCP client by configuring the mcpServers section accordingly (see example in the README).

Prerequisites recap: Python 3.12, uv (or pipx), and valid Google Ads credentials in google-ads.yaml.

Tips and common considerations:

• Environment variable: GOOGLE_ADS_CREDENTIALS can be used to point to your google-ads.yaml if your deployment requires explicit path configuration.
• If using Gemini, add the Google Ads MCP server under mcpServers with the appropriate command and arguments as shown in the README example.
• Timeouts: consider specifying timeout in your MCP client configuration (e.g., 30000 ms) to handle longer Google Ads API calls.
• Security: treat the google-ads.yaml file as sensitive information; limit access to the file and avoid exposing credentials in logs or configurations.
• If you encounter authentication issues, verify that your credentials file contains all required keys and that the login_customer_id is correctly set when needed.
• For development: you can run the server directly with uv and the module name ads_mcp.server to iterate on integration faster.