Required methodology for any task where the goal is to find the cheapest or best-value way to buy a product or service. Layered approach across retailers, inboxes, eBay, cashback portals, and credit cards. Skipping a layer means leaving money on the table.
Any task framed as one of the following triggers the deal-stack method automatically. No need to explicitly invoke it.
If the task is something else (just answer a question, summarize a doc, build a campaign), this rule doesn't apply.
Don't start at one store. Sweep the right retailers for the product category before committing to any of them.
| Category | Default retailers to check |
|---|---|
| Electronics / appliances / tools | Walmart, Amazon, Home Depot, Lowe's, Target, Best Buy, eBay, Costco · niche: AJ Madison, Town Appliance, B&H |
| Apparel | Amazon, Target, retailer-direct, eBay, Macy's, Kohl's |
| Groceries / household | Walmart, Target, Amazon Fresh, Fred Meyer / Kroger digital, Costco, Bi-Mart (Eugene) |
| Tools / construction supply | Home Depot, Lowe's, Harbor Freight, Northern Tool, Grainger, eBay |
| Services / SaaS | Direct vendor comparison + AppSumo / lifetime-deal sites |
Use Chrome MCP for retailer pages — never raw web search. The browser surfaces extension popups (Karma, Honey, Capital One Shopping, TopCashback, SimplyCodes, ID.me) that feed the rest of the stack. Capture what each extension offers as it fires.
Search both jason@bosstorque.ai (business) and jasonparkerjohnson@gmail.com (personal) for live promo codes from the retailers in scope. Pull actual code text out of email bodies — don't just note that an offer exists.
(from:RETAILER OR from:RETAILER_DOMAIN) (promo OR coupon OR "% off" OR "$ off" OR cashback OR rewards) after:2026/01/01Default scan window is the current calendar year unless the task specifies otherwise. Also check generic promo emails from:
Flag the expiration date on every code found. A code that expired last week is worse than no code at all — it sends the user down a dead path.
Always include eBay. Use the ebay-browse-api skill (the global default — see bt-ebay-api-access-may2026). Filter by exact name-brand SKU when there's an apples-to-apples comparison available with major retailers.
Sort by Price + Shipping ascending. Note seller feedback %, location, shipping speed, total price including any shipping cost. Flag international sellers as a risk (avoid unless Jason agrees).
When a same-SKU result on eBay beats a major retailer by $10+, surface it as a top contender — but explicitly flag the shipping delay tradeoff.
Pick exactly one portal per checkout — they're mutually exclusive. Capital One Shopping's browser extension auto-detects most major retailers and shows the rate inline. If the extension doesn't fire on a page, check the portal site directly.
| Merchant | Typical best portal (verify live) |
|---|---|
| Walmart | Capital One Shopping (~4%) or Karma (2%) |
| Amazon | Card-direct only (Amazon Visa 5%) — most portals don't cover Amazon |
| Home Depot / Lowe's | BeFrugal or TopCashback (1.5–2%) |
| Target | TopCashback (~2%) |
| Best Buy | TopCashback (~1%) |
| eBay | Rakuten or TopCashback (1–3%) |
| Wayfair / Macy's / Kohl's | Rakuten often wins |
Jason carries two cashback cards. Always add this as the last layer of the stack unless he says otherwise.
Mark the rate "~2% TBD" until Jason confirms the exact percentage. Don't double-count if a portal already rebates a percentage — the portal flow uses the same card as tender. Card cashback is additive only on direct-checkout (no portal redirect) purchases.
Always present at least three paths so Jason picks the tradeoff. Don't assume the cheapest wins if it's slow, and don't assume the fastest wins if it's expensive.
| Path | Speed | Effective net | Why pick this |
|---|---|---|---|
| A — same-day pickup | Today | $X | Need it now |
| B — 1-2 day delivery | This week | $Y (usually best balance) | Can wait a day or two |
| C — 5-10 day delivery | Next week | $Z (rock bottom) | Lowest absolute price |
For each top-ranked option in the final recommendation:
If a verification step fails, demote that option and re-rank. Don't recommend something you only saw in a search summary.
The methodology applied to a single-room window air conditioner for Eugene 97402:
Compare against the Amazon Basics path:
Path B wins on pure cost ($123 vs $133), Path A wins on speed (today vs Friday). Both got documented in the speed-vs-price matrix and Jason picked the tradeoff. Without the deal-stack method, neither stacked price would have surfaced.
# [Product] — Best Deal Sweep, [DATE]
[1-line winner statement with effective net and key path note]
## Top 5 options sorted by effective net cost
| Rank | Product · Source | Sticker | Coupon | Portal | Card | Net | Speed |
## Active coupons discovered in Gmail
| Source | Account | Offer | Code | Expires | Applies? |
## eBay results (same-SKU matches when applicable)
[Seller credibility, location, shipping speed, total]
## Three decision paths
A. Today pickup
B. 1-2 day delivery (best balance)
C. 5-10 day delivery (lowest absolute)
## Stores checked & eliminated
[Out-of-stock or not-carried items — shows breadth of search]~2% TBD, subject to availability, verify code at checkout.−$15.00 with the source.If Jason explicitly says "just check Walmart" or "I already know I'm buying from Amazon, just find the cheapest there" — skip layers 1-3 and run the stacking math (4-6) on the constrained set. Otherwise, run the full stack every time. Don't shortcut on assumption.
The Gmail scan (both accounts), the eBay check, the credit card layer, and the verification step. These four are where the actual savings show up. Skipping any of them means the recommendation is incomplete and Jason is leaving money on the table.
~/Documents/.claude/skills/deal-stack/SKILL.md~/Documents/.claude/skills/ebay-browse-api/SKILL.md + hub doc~/Documents/.claude/CLAUDE.md — "Deal-Stack Method — Primary Rule"