MayaVihin7.0 personal-shopper

Research products, compare options, and find the perfect gift based on recipient and occasion.

install

source · Clone the upstream repo

git clone https://github.com/Aryanrai-007/MayaVihin7.0

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/Aryanrai-007/MayaVihin7.0 "$T" && mkdir -p ~/.claude/skills && cp -r "$T/.local/secondary_skills/personal-shopper" ~/.claude/skills/aryanrai-007-mayavihin7-0-personal-shopper && rm -rf "$T"

manifest: .local/secondary_skills/personal-shopper/SKILL.md

source content

Personal Shopper & Gift Finder

Research products, validate prices/reviews, and generate gift ideas that aren't generic.

When to Use

"What's the best [X] under $[Y]?" / product comparison
"Is this Amazon deal real?" / price validation
Gift ideas for a specific person

When NOT to Use

Market research (deep-research), budgeting (budget-planner), used cars (used-car-advisor)

Research Sources — Where to Actually Look

Review & Research Sites

Category	Best source	Why
Most consumer goods	Wirecutter (nytimes.com/wirecutter)	Long-term testing, updates picks when they fail
TVs, monitors, headphones, soundbars	`rtings.com`	Lab-measured data (input lag in ms, frequency response graphs), not vibes
Appliances, cars, mattresses	Consumer Reports (paywalled) — search `"consumer reports [product] reddit"` for summaries
Enthusiast gear (knives, keyboards, flashlights, coffee, pens)	Product subreddit wiki/FAQ — `site:reddit.com/r/[hobby] wiki`	Actual users, not affiliate sites
Outdoor/camping	`outdoorgearlab.com`	Side-by-side field testing
Laptops	`notebookcheck.net`	Thermals, throttling, display calibration data
Skincare/cosmetics ingredients	`incidecoder.com`	Ingredient breakdown, no marketing

Curated & Boutique Sources

Prefer these over generic Amazon results — they surface more interesting, unique finds:

Source	Best for	Why
Wirecutter (nytimes.com/wirecutter)	Everyday products, gift guides	Rigorously tested, regularly updated
Conde Nast Traveler / GQ / Bon Appetit	Travel gear, fashion, food/kitchen	Editorially curated, taste-driven
Goop	Wellness, beauty, home, unique gifts	Curated luxury, discovers interesting small brands
Strategist (nymag.com/strategist)	Gift guides, home, fashion, wellness	Real-person recommendations, not algorithm-driven
Cool Material ( `coolmaterial.com` )	Men's gifts, gear, home goods	Curated interesting finds
Uncommon Goods ( `uncommongoods.com` )	Unique/artisan gifts	Handmade, small-batch, creative
Food52 ( `food52.com` )	Kitchen, home, food gifts	Chef-tested, beautifully curated
Reddit gift threads	Any category	Search `site:reddit.com "[category] gift"` or `"best [product] reddit"` — real opinions from enthusiasts

Search pattern for honest reviews:

"[product] reddit"

"[product] site:reddit.com"

— cuts through SEO affiliate spam. Also

"[product] long term"

"[product] after 1 year"

Search pattern for curated finds:

"[product/category] site:nymag.com/strategist"

"best [category] gifts site:goop.com"

— surfaces editorially picked items over algorithm-promoted ones.

Price Validation — "Is This Deal Real?"

Amazon "40% off" is often off a fake inflated list price. Verify:

Tool	Use	Access
CamelCamelCamel	Amazon price history chart — paste URL or ASIN	`camelcamelcamel.com` (free, webFetch works)
Keepa	Same but overlays directly on Amazon pages; more marketplaces	`keepa.com` (free tier sufficient)

Read the chart: if "sale" price = the price it's been at for 6 of the last 12 months, it's not a sale. Real deals sit at or near the all-time low line. Flag any product where price spiked up right before the "discount."

Fake review detection: Fakespot shut down July 2025; ReviewMeta is currently down. Manual heuristics:

Cluster of 5-star reviews in a 2-day window = paid review burst
Reviews that mention "gift" / "haven't tried yet but looks great" = incentivized
All reviews are 5 or 1 stars, nothing in between = manipulated
Check reviewer profiles — dozens of 5-star reviews across random categories = fake account
Sort by most recent, not "top" — recent reviews reveal quality decline after a product gets popular

Product Recommendation Format

Always give 3 tiers so the user can self-select on budget:

Budget pick — 80% of the performance at 40% of the price
Best overall — the Wirecutter-style default
Upgrade — only if the premium is justified by a specific use case; say what that use case is

For each: price, one-line "why this one," one-line "main tradeoff," and always include direct links:

Product link — link to where the user can actually buy it (Amazon, retailer site, etc.). Search for the specific product and provide the real URL, not a homepage.
Review/source link — link to the review, article, or Reddit thread that informed the recommendation
Price history link — for Amazon products, include a CamelCamelCamel link so the user can check price history themselves

Never recommend a product without at least a purchase link. The whole point of a personal shopper is saving the user time — making them search for the product themselves defeats the purpose. Use webSearch to find actual product pages and verify URLs are live before sharing.

Gift Framework — Beyond "Know the Person"

The four gift modes (pick one, don't blend):

Upgraded everyday — a nicer version of something they use daily but would never splurge on (good olive oil, merino socks, quality umbrella). Safest bet. Works for anyone.
Experience — class, tickets, tasting, subscription. No clutter. Good for people who "have everything."
Consumable luxury — fancy food/drink/candle they'll use up. Zero storage burden. Default for acquaintances, hosts, coworkers.
Interest-deep-cut — something only a real enthusiast would know about. Highest risk, highest reward. Requires research: search
```
r/[their hobby] "gift"
```
or
```
"best gifts for [hobby] enthusiast reddit"
```
.

Extraction questions (ask user, not recipient):

What do they complain about? (Complaints → unmet needs → gifts)
What have they mentioned wanting but not bought? (The $80 thing they keep not pulling the trigger on)
What do they already own a lot of? (Signals the interest; buy adjacent, not duplicate)
What did they get excited about recently?

Variety rule — this is critical:

Recommendations must span different categories. If someone asks for a gift, don't suggest 3 fragrances or 3 candles or 3 books — spread across different types of products unless the user specifically asked for a single category. For example, a good gift list might include one kitchen item, one experience, and one piece of gear. Variety shows thoughtfulness; a list of same-category items shows laziness.

Hard rules:

Scented anything (candles, perfume, lotion) — only if you know their taste. Scent is personal.
No decor unless you've seen their space
No clothing with sizes unless you're certain
Gift receipt always. Return window matters more than wrapping.

Occasion	Default mode	Budget anchor
Close friend birthday	Interest-deep-cut or upgraded-everyday	Whatever you'd spend on dinner together
Acquaintance / coworker	Consumable luxury	$20-40
Housewarming	Consumable (nice pantry goods, wine) — no decor	$25-50
Wedding	Registry. If off-registry, cash.	Cover your plate cost minimum
Thank-you	Consumable, handwritten note matters more than price	$15-30
Host gift	Something they can use after you leave (not flowers — requires a vase and attention mid-hosting)	$15-30

Gift recommendations must also include direct purchase links. For each gift idea, provide a link to a specific product the user can buy — not just "nice olive oil" but a link to a specific bottle on a specific site.

Limitations

Can't see real-time stock/price — always tell user to verify before buying
Can't access paywalled review sites directly (CR, some Wirecutter)
Can't process transactions