MCP recipe
Answer features scan
See how often each engine’s answers used web search or held shopping results, local businesses, ads, video, images or tables.
The prompt
Use query_metrics, broken down by engine, over a 28-day window. A query takes at most five metrics, so run two queries: first web_search_rate, shopping_rate, local_businesses_rate and ads_rate, then video_rate, images_rate and tables_rate. Put the results in one table with engines as rows and the seven rates as columns, each with its observations. A null rate means that engine does not report the feature or no answer was checked; show it as a dash, never as 0%. Flag provisional values.
Paste it into an assistant connected to the DiscoveredBy MCP connector. The key needs the scope: explorer:read.
Tools it calls
query_metrics (explorer:read)
Computes Explorer metrics for the project now, the same numbers the app’s Explorer shows.