Search Term Report vs Cerebro Ranks with Claude

Here are my keyword performance export and targets: target ACOS [X]%, harvest after [N] orders, negate after [M] clicks with 0 orders.
List search terms to promote to exact match and terms to add as negatives, with the campaign and ad group for each, as a Sponsored Products bulk file I can review before upload.

Before you recommend any change to my Amazon campaigns, tell me: what in this data supports it, what could go wrong, and how I'd know within 14 days whether it worked.
Recommend at most 2 changes at a time, never touch my top-selling SKUs' existing campaigns, and remind me of the baseline numbers so I can roll back.

Review my Sponsored Products targets for the last 30 days. For each target with ACOS above [X]% or more than [N] clicks and no sales, recommend lower bid, pause or keep, with the reason.
Show the list first; only make the changes I approve.

Here are my Amazon DSP exports for the last 4 weeks: campaign report, reach and frequency, and device breakdown.
1. Show average frequency per line item by market and flag any above 7.
2. Show spend share and ROAS by device type and suggest bid-modifier changes.
3. List the 5 changes you'd make before the next flight, each with the number that supports it. Don't assume data that isn't in the files.

Act as my Amazon launch operator for a new SKU. Step by step, and wait for my approval between steps:
1. From the keyword and competitor data I paste below, pick the 15 highest-intent keywords and explain why.
2. Write a title, 5 bullets and backend search terms using those keywords. No claims I haven't given you.
3. Propose a launch campaign structure (Auto, a Discovery/broad campaign, and an exact-match Sales campaign) with starting bids and daily budgets calibrated to the CPCs in the data.
Product details: [paste]
Keyword + competitor data: [paste]

Here is my search term report for the last [14] days [attach].
Give me, in plain English: the search terms wasting money (spend over $[x] with no orders) and the ones worth scaling (orders with ACoS under [y]%).
For each, the exact change you'd make. Don't make any change until I confirm it.