ZooData — Amazon Market Analysis

Respond in the user's language.

Start here

1. Read and apply references/analysis-contract.md before classifying the request, making an evidence call, or sending user-facing text.
2. Classify the current question. Route an open-ended "what market/product should I explore?" request to discover; a named category or niche and entry/competition question to evaluate; a change, trend, or watch question to track. A single field or one dated snapshot is a quick lookup. A product/ASIN may support a market question, but a listing audit, competitor war room, or price-setting request belongs to its dedicated skill.
3. Read references/cli-contract.md and references/reference.md; inspect the selected bundled CLI subcommand's --help. The reference owns production endpoint availability, parameters, response shapes, dates, and credit facts.
4. Load references/output-rules.md. For a quick lookup, load only references/execution-guide.md sections Authority and routing, Execution mode, Field Identity Gate, Interface Failure Stop Gate, Final Output Gate, and Credential and Credit Failures.
5. For a full analysis, load the complete references/execution-guide.md, references/evidence-protocols.md, and exactly the applicable scenario: references/scenarios-discover.md, references/scenarios-evaluate.md, or references/scenarios-track.md. The guide owns market stage selection and domain Gate decisions; the scenario owns its evidence and conclusion scope. Reclassify a follow-up from the user's actual question rather than forcing the next story step.
6. Before interpreting returned market measures, load references/market-metric-semantics.md. Before interpreting seller capital, cost, margin, risk tolerance, or compliance claims, also load references/seller-input-semantics.md.

Source-of-truth boundaries