About Mcp Google Ads
Mcp Google Ads is a multi-account Google Ads MCP server for operators, allowing client or in-house accounts to be switched without restarting. It provides tools for account, campaign, keyword, search-term, and ad-performance analysis.
Key features
- 01Route across accounts and set a default
- 02Discover accessible customer accounts
- 03Summarize spend and compare periods
- 04Query campaigns and performance metrics
- 05Analyze keywords, search terms, and ads
- 06Retry eligible API calls
Use cases
- 01Review ad data across client accounts
- 02Compare account spend across time periods
- 03Analyze keyword and search-term performance
- 04Inspect campaign and ad results
MCP support
The server exposes Google Ads reporting and account-routing capabilities as tools to compatible MCP clients, using GAQL to query the Google Ads API.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
uvx mcp-google-ads-multiGOOGLE_ADS_DEVELOPER_TOKENSecretGoogle Ads API developer token from the Google Ads API CenterGOOGLE_ADS_ACCOUNTS_CONFIGPath to accounts.json config file with OAuth credentials per accountLocal servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
10Saved tool list · updated 2w ago
This tool takes no parameters.
Details
From the project READMEmcp-google-ads
Multi-account Google Ads MCP server for operators who need one MCP surface across multiple client or in-house ad accounts without restarting the server.
Release posture: beta package, version 0.1.0 from pyproject.toml.
Choose your path
| You are... | Start here | Then |
|---|---|---|
| Connecting the server to your MCP client | docs/start-here.md | Quick start below |
| Auditing account switching and retries | docs/architecture.md | gads/server.py |
| Reviewing packaging metadata | pyproject.toml | accounts.example.json |
Architecture
flowchart TD
U[AI operator] --> C[MCP client]
C --> S[FastMCP server]
S --> M[Account manager]
M --> F[Accounts config]
S --> Q[GAQL query helpers]
Q --> R[Retry wrapper]
R --> A[Google Ads API]
A --> R --> S
Request flow
flowchart TD
P[Operator asks for account or campaign data] --> T[Selected MCP tool]
T --> A[Resolve named or default account]
A --> B[Build Google Ads client]
B --> C[Run GAQL query]
C --> D{API call succeeds?}
D -- yes --> E[Normalize rows and return]
D -- retryable --> F[Backoff and retry]
F --> D
D -- no --> G[Return error payload]
Quick start
- Install the package.
python -m pip install mcp-google-ads-multi
- Copy the accounts config template and add your accounts.
mkdir -p ~/.config/mcp-google-ads
cp accounts.example.json ~/.config/mcp-google-ads/accounts.json
- Register it in your MCP client.
{
"mcpServers": {
"google-ads": {
"command": "uvx",
"args": ["mcp-google-ads-multi"],
"env": {
"GOOGLE_ADS_DEVELOPER_TOKEN": "your-developer-token",
"GOOGLE_ADS_ACCOUNTS_CONFIG": "/Users/you/.config/mcp-google-ads/accounts.json"
}
}
}
}
Available tools
| Tool group | Tools | Purpose |
|---|---|---|
| Account selection | list_accounts, set_default_account, list_customers | Manage account routing and discover accessible customers |
| Spend summary | get_account_summary, compare_periods | Period-over-period account reporting |
| Campaign analysis | list_campaigns, get_campaign_performance | Campaign inventory and range-based metrics |
| Query-level detail | get_keyword_performance, search_terms_report, get_ad_performance | Keyword, search-term, and ad reporting |
Runtime proof
| Claim | Proof |
|---|---|
| Package entry point is stable | mcp-google-ads-multi = "gads.server:main" in pyproject.toml |
| Multi-account routing is first-class | AccountManager() and set_default_account() in gads/server.py |
| Queries go through reusable helpers | gads/query.py and gads/retry.py |
| Config is file-driven | accounts.example.json documents the accounts shape |
Repo map
| Path | Purpose |
|---|---|
gads/server.py | FastMCP tool surface |
gads/accounts.py | Account config, OAuth/service-account loading |
gads/query.py | GAQL execution helpers |
gads/retry.py | Retry behavior for API calls |
docs/start-here.md | Setup, validation, common failures |
docs/architecture.md | Component map and request lifecycle |
Validation
| Check | Command |
|---|---|
| Import compiles | python -m compileall gads |
| Package builds | python -m build |
| README/docs links are local | rg '\\]\\(([^)]+\\.md)\\)' README.md docs/ |
License
MIT
Project context for new tasks
Use INDEX.md to find task-specific contracts; consult prompt.md when product direction is relevant. Historical examples do not select the current task.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.