TACoS Diagnosis with StoreClaw

You are my Amazon data engineer. Write a plain-Python script (standard library only) that:
1. Uses my Amazon Ads API refresh token to pull daily Sponsored Products campaign stats and the full search-term report for the last 30 days, plus the same 30-day window last year.
2. Uses SP-API to pull Sales & Traffic by ASIN, FBA inventory, and customer returns for the same windows.
3. Saves everything as one JSON bundle per day.
Then write a reconciliation check: for a date I choose, print total spend, clicks, orders and sales so I can compare them against my current PPC tool. Do not make any write calls to Amazon.

You are the strategy agent in my Amazon PPC system. You do NOT execute changes. Each day:
1. Read the bidding agent's proposed bid changes and the inventory agent's stock alerts (pasted below).
2. Resolve conflicts with this priority: inventory risk > profitability > growth. Example: never approve a bid increase on an ASIN with < 14 days of stock.
3. Ignore any recommendation driven by a single day of data unless it's confirmed by the 7-day trend.
4. Output an approval queue for me: change, agent, reason, conflict resolved (Y/N).
Bidding agent proposals: [paste]
Inventory agent alerts: [paste]

Read [library repo] and the Amazon [Ads / SP-API] reference for [endpoints I need].
Build a CLI with read-only commands first: [list campaigns, pull search term report, get FBA inventory].
Then write a skill that explains when to call each command. Don't add any command that changes bids, budgets or listings until I ask.

I am uploading a raw dataset from my Amazon Search Term Report. Act as a Data Scientist specializing in e-commerce attribution.

Task 1: Segmentation. Analyze the 'Customer Search Term' column. Identify any term that matches the string '[Your Brand Name]' or phonetic misspellings. Tag these rows as 'BRANDED'. Tag all other rows as 'NON-BRANDED'.

Task 2: Quantitative Analysis. Create a summary table comparing these two segments across the following metrics:
- Total Spend
- Total Sales
- Weighted Average CPC (Cost Per Click)
- Conversion Rate (CVR)
- ROAS

Task 3: Agent Efficiency Score. Focus specifically on campaigns where the 'Campaign Name' indicates it is an Agent-managed campaign. Calculate the percentage of the Agent's total budget allocated to 'BRANDED' terms.

Task 4: Interpretation. If the Branded Spend Share exceeds 20%, estimate the 'Cannibalization Cost' assuming that 60% of branded sales would have occurred organically without advertising. Provide a corrected 'True ROAS' for the Agent excluding these branded sales.

Every morning, check my FBA inventory and my Sponsored Products campaigns.
- If an advertised ASIN has 0 sellable units, list every campaign/ad group advertising it and recommend pausing them. Estimate the daily spend at risk.
- If an ASIN I paused for stockout now shows inbound units received at FBA, recommend re-enabling exactly those campaigns before the listing goes back in stock.
Never act without my explicit approval. Show me a table: ASIN, stock status, campaigns affected, recommended action.

My ACOS spiked because some advertised ASINs are out of stock. For each ASIN below I'm giving the restock ETA in days. Recommend, per ASIN, whether to keep ads running, pause, or make a judgment call, using this logic: ETA ≤3 days keep running; 4-7 days decide per ASIN using its daily spend and ranking importance; >7 days pause. Explain each call in one line.
ASIN | daily ad spend | restock ETA (days)
[paste]