Sponsored Products campaign auditor

Purpose

Produce a read-only, portfolio-level audit from Sponsored Products campaign data obtained through Amazon Ads MCP. Measure structure and delivery, explain uncertainty, and identify review candidates. Do not create, update, pause, archive, or otherwise mutate campaigns.

Account-context recovery

If Amazon account scope, identifier type, marketplace mapping, or account relationships become unclear, consult amazon-ads-accounts when it is available. Resume this skill after resolving the ambiguity. If it is unavailable, use equivalent read-only discovery and ask the user when multiple valid choices remain. Never guess or interchange identifier types.

Workflow

1. Inspect the connection. Confirm that the configured Amazon Ads MCP exposes the account-discovery and reporting operations needed for this audit. Use the actual exposed tool names and schemas; do not invent aliases. Read the live-reporting reference before requesting data.
2. Resolve scope. Establish the identity, advertiser account/profile, region, marketplace, and currency. Confirm the date range and the questions the audit should answer. Ask only for choices that materially change extraction or interpretation.
3. Define the report. Discover the current report-field catalog, choose Sponsored Products campaign dimensions and metrics, and validate the complete field set before creating the report. Record the exact accepted field IDs.
4. Retrieve safely. Create the report and retain its report ID. Follow the MCP's asynchronous status contract. Avoid tight polling; when the client cannot wait safely, return the report ID and resume retrieval when the user continues.
5. Establish grain. Inspect the returned rows before calculating metrics. Aggregate campaign-day rows to campaigns; sum additive metrics and treat budget according to its type and time grain.
6. Apply the audit. Read the audit rules. Use user-provided targets or operating standards when available. Without them, report distributions and review candidates rather than declaring a universal ideal.
7. Deliver the result. Follow the report template. Mark unsupported sections N/A with the missing field or ambiguity, instead of estimating absent values.