MCP recipe
Prompt performance review
Find the prompts where your brand gained or lost the most visibility over the last 28 days.
The prompt
Use list_prompts to read every current prompt, following next_after_id until it is null, and note which ones are paused (active is false). Then use query_metrics with brand_visibility and answers_analysed, broken down by prompt, over a 28-day window compared with the previous period. List the five prompts with the largest gain and the five with the largest drop in brand visibility, each with its prompt text, value, prior value and observations. Mark paused prompts, flag provisional values, and say whether the result reports the prompt breakdown as truncated.
Paste it into an assistant connected to the DiscoveredBy MCP connector. The key needs these scopes: prompts:read, explorer:read.
Tools it calls
list_prompts (prompts:read)
Current prompts, including paused prompts and their current labels.
query_metrics (explorer:read)
Computes Explorer metrics for the project now, the same numbers the app’s Explorer shows.