Field guide
A printable explanation of MCP, Xnurta, and the five checks, with examples you can use at work.
Open the PDF guide ->Learn what MCP does in plain English. Use one Amazon ads question to practice asking clearly, finding the right information, checking it, and choosing a next step.
Read the guide, look at the maps, or watch the short lesson. They all use the same simple example.
A printable explanation of MCP, Xnurta, and the five checks, with examples you can use at work.
Open the PDF guide ->Four pictures show who does what, where the information comes from, and where Xnurta fits.
See the diagrams ->A short voice-led lesson follows one question from the first ask to a next step another person can review.
Watch the explainer ->Start with a question, set its limits, find the facts, check them, and decide what to do next.
MCP is a shared way for an AI app to ask another service for information. The app you talk to is the host, the connection is the client, the service is the server, and the original system still owns the record.
Remember this: MCP helps an AI app reach a service. It does not become the owner of that service's data.
For example, Xnurta can provide advertising numbers while a stock or finance system provides the rest of the business picture. Keep those jobs separate so the answer is easy to check.
Fill in the four things a teammate needs to ask the same question: what you are looking at, dates, numbers, and answer format.
Question: Why did our practice campaign's return from ads fall week over week? Account: one approved Amazon ads account Dates: May 5-11, 2025 vs Apr 28-May 4, 2025 Numbers: return from ads, spend, sales, orders Answer: short table + what we still need to check
Clear questions have limits. Asking in plain English does not give an AI app access to every account. Say which account, dates, and decision you mean.
Open the question lessonWrite down the return from ads, spend, sales, orders, source, dates, and how current the data is before you suggest anything.
Use it to find and organize information. Keep the source, the limits, the uncertainty, and the person responsible visible.