Top 10 Xnurta MCP prompts for Amazon Ads, with the data each needs

What do the ten prompts need at a glance?

Each prompt needs a connected ad account, and the last three also need write access. Xnurta MCP connects Claude, ChatGPT and other MCP clients to your Xnurta data, so the prompts are plain language and you swap the [placeholders] for your own account, campaign and numbers. If you have not connected yet, Connect Claude or ChatGPT to Xnurta MCP walks through it, starting from the endpoint https://mcp.xnurta.com/mcp.

Write access comes in three kinds, each with its own permission scope: managed group management, ad-entity editing and SP campaign creation. The table shows which prompt needs which.

How do you phrase a prompt so the answer is usable?

Name the object, give a time window and a comparison, and say what output you want. That is most of it. "My campaigns, recently" gets you a shrug. "Campaign X, last 7 days against the prior 7, as a table" gets you a decision.

Two more habits matter. Ask follow-ups in the same chat: start broad, then ask why, then drill into one campaign. And before any write, ask the assistant to list the objects and the exact values it plans to change. Not sure which ad accounts you have? Ask it to list the ones you are authorized to access, then pick one. That is Prompt 1.

How far back does the data go, and how fresh is it?

About 15 months, across 22 marketplaces. Performance data is daily by default. Hourly is available for SP, SB and SD campaigns and for SP keywords, and the query window should stay within about one week. Freshness follows the platform's update cadence, so a number is current to the last refresh, not to the second.

You see only the ad accounts your own Xnurta login can access, main accounts and sub-accounts included.

Which prompts show you where an account needs attention?

Prompts 1 to 3 do. They start broad, find the campaign that is off, then look at the search terms behind it. All three are read only.

List all ad accounts I am authorized to access. How did each one perform yesterday? Identify the five issues that need the most attention and support each one with data.

Campaign [campaign name] in ad account [account name] saw ACOS rise from [X]% to [Y]% over the last 7 days. Analyze the causes across budget, bids, targeting and structure. Recommend specific actions to lower ACOS and increase orders, and show the numbers behind each cause.

Which search terms performed well in ad account [account name] over the last [30] days? Show CTR, CVR, ACOS and orders. Which terms deserve more investment, and which high-ACOS, low-conversion terms should be negated? Do not change anything.

Prompt 2 pulls performance and the campaign's current settings together, which is why one question can cover budget, bids, targeting and structure. Prompt 3 returns a list of negatives. Applying that list is an ad-entity edit, so it follows the access rule below, and you should ask the assistant to list the exact negatives and wait for your confirmation before it executes.

Which prompts tell you what changed, and what it means?

Prompts 4 to 7 do. Prompt 4 reads the operation log, which holds human and AI actions on campaigns, ad groups, targets, bids, budgets and managed groups. You can filter by operator, so you can ask for human actions and AI actions separately. Prompt 5 is the only one on this list that needs a file from you.

Who changed campaign [campaign name]'s budget in ad account [account name], and when? List all human actions from the last [30] days. Then list the automatic bid changes and negative-targeting actions AI performed in that account over the last [7] days.

Here is the unit margin for each ASIN in product line [product line name]: [paste or attach your data]. Based on actual margin, which campaigns are losing money, and should I cut them or increase their budgets? Show the math and the date range you used.

Compare spend share and ACOS across SP, SB, SD and match types in ad account [account name] for the last [30] days. Recommend how to reallocate the budget. Do not change anything.

Create a weekly advertising report for ad account [account name] covering last week: period-over-period KPIs, a 7-day trend, top movers, and recommended actions for next week.

Prompt 5 is the one the platform alone cannot answer. Ad data does not know your unit margin, so you hand it over and the assistant combines it with your advertising data. Cross-analyze Amazon Ads with your own data in one prompt covers the file side. For Prompt 4 in depth, read Who changed my Amazon campaign?.

Prompt 7 also has an official skill behind it. With the skills installed, asking for last week's ads report triggers the Weekly Ads Report skill, which is built to return week-over-week KPIs, 7-day trends, anomalies, top movers and next-week suggestions. The same family covers Monthly Ads Report, Product Diagnosis, Ads Structure Analysis and Search Term Traffic Diagnosis. More in How an SOP becomes something an AI assistant can run.

What do you need before a prompt can change your account?

You need write access switched on for your account, and a permission scope that matches the kind of change. Reading never needs it. Queries on performance, settings and operation logs are read only.

Each kind of write is separate. A connection that can edit ads cannot necessarily create campaigns, so a failed Prompt 10 on a connection that handled Prompt 8 is expected, not broken. For more on read and write access, see Read-only vs read-write MCP for Amazon advertisers.

Which three prompts change something, and what do they ask first?

Prompts 8 to 10 do, and each one tells the assistant to show you the change before it makes it. One uses ad-entity editing, one uses managed group management, one uses SP campaign creation.

Which campaigns in ad account [account name] hit their budget cap in the last 7 days? Keep only those with ACOS below [X]%. Raise each daily budget by [10]%. Show every campaign with its current budget and its new budget, and wait for my confirmation before you submit anything.

Find campaigns in ad account [account name] that had spend in the last 30 days but are not under AI management, and group them by how close their ACOS values are. Create managed groups from those groups of SP campaigns. Set target ACOS to [6]% and leave AI switched off for now. Before you execute, show me the groups, target ACOS, AI status, action space and campaign list.

Create a paused SP auto campaign for these [5] ASINs: [ASIN list]. Set a daily budget of [amount] and show me the full structure before you submit it.

Prompt 8 touches money, so the assistant returns a preview, which objects and from what value to what value, and submits only after you confirm. Prompt 9 builds managed groups with AI switched off, which lets you check the groups before the AI starts working on them. Deleting a managed group later archives it, and archiving cannot be undone. For the levers a group exposes, see Read and change Xnurta's AI levers through MCP.

Prompt 10 creates the campaign in the state you ask for, so ask for paused. Creation is all-or-nothing: if one campaign in a request fails validation, none are created. Campaigns cannot be deleted afterward, only archived, and archiving is irreversible. The full walkthrough is Create an Amazon campaign from a launch brief, and Edit Amazon campaigns in bulk through an AI assistant covers edits across many campaigns.

Which prompt should you run first?

Run Prompt 1, then Prompt 5. Both are read only, and together they tell you which ad accounts the assistant can see and which campaigns are losing money at true margin. Then pick one write prompt and run it on one account.

Save the ones you repeat. A skill teaches the assistant the routine for a task, and you can build your own and keep it private or share it with your organization. The Weekly Ads Report is the model: Prompt 7 turned into a routine your whole team runs the same way.