| name | shopping-hunter |
| description | Deep shopping research skill for finding products similar to a reference link or image, benchmarking global pricing, comparing quality/provenance, and creating a reusable buying dossier with links, per-item prices, shipping notes, and value-focused recommendations. |
Shopping Hunter
Use this skill when the user wants to find, benchmark, compare, or source products similar to a reference item, image, aesthetic, material, maker, or buying goal. The default mode is deep global shopping research with Australia prioritized, AUD-normalized pricing, and a documented dossier.
Default operating mode
- Produce all of the following:
- A buying dossier.
- A concise shortlist of the best buys.
- A market benchmark showing cheap, mid-market, premium, and high-end pricing.
- Prioritize Australia first, then global sources.
- Normalize prices to AUD primary, while preserving original currency and set size.
- Be value-focused: call out bargains, fair pricing, overpricing, provenance premiums, hidden shipping risk, and “cheap but not equivalent” alternatives.
- Prefer actual purchasable product links over inspiration-only pages.
- Include official brand/maker sites, marketplaces, local retailers, boutiques/design stores, secondhand/vintage listings, high-end benchmarks, and wholesale/B2B leads when relevant.
- Search deeply: use regional/source passes, price-tier passes, terminology/source-origin research, exact-source tracing, and high-end benchmarks.
Required intake questions
Before starting, ask for the missing items from this list. If the user already provided some, only ask for the missing/high-impact ones.
- Reference link(s) and/or image(s).
- What they like about the reference: colours, shape, material, finish, dimensions, provenance, vibe, brand, price, etc.
- Budget or price bands to benchmark.
- Must-have traits vs nice-to-have traits.
- Regions to prioritize, with Australia as the default priority.
- Whether vintage/used/one-off listings are acceptable.
- Whether wholesale/B2B/MOQ sources are acceptable.
- Quantity needed and whether sets are preferred.
- Deal-breakers: materials, countries, shipping, lead times, ethics, returns, fragility, defects.
- Whether to create a Markdown dossier file; default is yes.
If the user says “don’t hold back,” “deep search,” or similar, proceed with a deep multi-pass workflow and do not over-ask beyond truly missing constraints.
Research workflow
1. Parse the target
Extract a target profile:
- Product category and functional use.
- Visual traits.
- Materials and construction.
- Dimensions/capacity if relevant.
- Likely origin/provenance clues.
- Search vocabulary and synonyms.
- Known brand/source clues from the reference page.
Separate:
- Exact match: same brand/product/source.
- Close match: same material/process/aesthetic, different maker.
- Adjacent: same vibe, different process or quality.
- Budget lookalike: visually similar but materially different.
- Premium benchmark: higher-end equivalent for price context.
2. Run parallel research passes
When available, use parallel subagents or independent searches by scope. Good default passes:
- Exact-source tracing: same product, same brand, resellers, wholesale clues.
- Australia/NZ retailers.
- Etsy/global marketplace alternatives.
- Official brand/maker sites.
- US homeware/boutique sources.
- UK/EU homeware/boutique sources.
- Budget marketplace and mass-market alternatives.
- Wholesale/B2B/source-origin pass.
- High-end/designer/artisan benchmark.
- Search-terms/provenance map.
Do not duplicate work already delegated to an agent. Wait for required research before final synthesis when feasible.
3. Verify and normalize
For each candidate, capture:
- Product name.
- Merchant/seller.
- URL.
- Price as listed.
- Currency.
- Set size / quantity.
- Calculated per-item price.
- Approximate AUD per item.
- Shipping to Australia or availability notes.
- Materials and process.
- Origin/provenance.
- Dimensions/capacity.
- Similarity rating.
- Recommendation tier.
- Risk notes: blocked page, unverified price, vintage defect, MOQ, sold out, surface coating vs material colour, import/customs risk.
Never invent prices. If a price is from a snippet, archive, search result, or blocked page, label it clearly as unverified or partially verified.
4. Build the dossier
If the user wants a file, create a Markdown dossier in the session files folder unless they explicitly request a repository path. If the user explicitly asks for a repo file, create it where requested.
Recommended dossier structure:
- Executive verdict.
- Price bands / market map.
- Best shortlist.
- Exact-source findings.
- Local/Australia options.
- Global alternatives by region.
- Budget/mass-market benchmarks.
- Premium/high-end benchmarks.
- Wholesale/source-origin leads.
- Search terms and sourcing vocabulary.
- Buying guidance.
- Research limitations and unverifiable leads.
Use dense tables for product comparisons. Keep a short chat summary after creating the dossier and include the file path.
Recommendation style
Be direct and opinionated:
- Say which option is the best exact match.
- Say which option is the best value.
- Say which option is cheapest but materially different.
- Say which option is premium and why.
- Say whether the reference price looks fair, inflated, or unusually cheap.
- Explain when provenance, material, handmade process, or shipping justifies a higher price.
Quality rules
- Prefer current, live product pages, but include archived/snippet data when clearly labeled.
- Include sold-out/vintage listings only when useful for benchmarking or search vocabulary.
- Distinguish colour-in-glass from sprayed/painted/electroplated finishes.
- Distinguish handmade, mouth-blown, machine-made, pressed, molded, and studio-art products.
- Distinguish retail price from landed cost; shipping can dominate international purchases.
- Flag fragile/shipping-sensitive categories.
- Do not treat marketplace SEO claims as proven provenance without corroborating text.
Final response pattern
Lead with the outcome:
- “Created the dossier:
<path>”
- “Best exact match: ...”
- “Best value: ...”
- “Price verdict: ...”
Keep the chat response concise; the dossier carries the detail.
