mcpField
S
shop-amazon

shop-amazon

github.com/jlave-devProductivity256 installs
Repository

About shop-amazon

A shopping research skill that searches Amazon and verifies candidate listings on their product pages before returning ranked recommendations. It gathers shopping constraints first and treats account actions such as purchases as off-limits unless explicitly requested.

Key features

  1. 01Search Amazon listings
  2. 02Verify title, price, rating, and review count
  3. 03Rank products by stated criteria
  4. 04Include product trade-offs
  5. 05Keep shopping research read-only by default

Use cases

  1. 01Compare products within a budget
  2. 02Find an item for a specified use or gift
  3. 03Check product details before choosing
  4. 04Review verified options and trade-offs

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/jlave-dev/agent-skills --skill shop-amazon

After installing, type/shop-amazonin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

agents/
openai.yaml
references/
asin-extraction.md
common-errors.md
output-formats.md

Details

From the project README

Shop Amazon

Search Amazon, verify product facts on product pages, and return ranked recommendations.

Requirements Gate

Before browser automation, ask directly for any missing shopping constraints:

  • Budget or comfortable price range
  • Usage context: personal, gift, professional, compatibility need, etc.
  • Deal-breakers: required features, avoided brands/materials, shipping needs

If the user already supplied enough constraints, proceed.

Browser Access

Open Amazon through the normal browser surface available in the current environment. Prefer Chrome when that is how Codex usually interacts with web pages here; if another browser or page-interaction path is available, use that.

Keep the same browser page or session for search and product verification when possible. Leave Amazon open only when the user explicitly asks for it.

Workflow

  1. Search Amazon directly at https://www.amazon.com/s?k=<encoded search query>.
  2. Extract product names and ASINs from the same result container. Never pair a list of names with a separate list of ASINs.
  3. Visit each candidate product page before presenting it.
  4. Keep only products whose product page confirms title match, current price, rating, and review count.
  5. Rank by the user's criteria. When ratings are within 0.5 stars, prefer the product with more reviews.

Read references only when needed:

Account Actions

Do not add items to cart, buy, reorder, subscribe, send gifts, apply coupons, or start checkout unless the user explicitly asks for that exact action. Treat product research as read-only by default.

Output Rules

  • Present only product-page-verified items.
  • Use clean links: https://www.amazon.com/dp/<ASIN>.
  • Include ASIN, product-page price, rating, review count, why it matches, and key trade-offs.
  • Mark each item (verified) only after product-page title, price, rating, and review count were checked.

Comments

Markdown supported