Skip to main content
Glama
shopping-mcp
by Nuu-maan

shopping-mcp

MCP server that looks up live listings on Indian storefronts so an agent can quote real prices, stock, and specs instead of guessing.

Currently wired to Flipkart and Amazon.in. The layout is store-agnostic so more marketplaces can be added the same way.

What it does

Tool

Store

Returns

autosuggest

Flipkart

Query completions

search_products

Flipkart

pid, URL, selling price, MRP, rating, stock, key specs

list_search_filters

Flipkart

Facets (extra_facets) for the current query

get_product

Flipkart

Price, offers, highlights, specs, similar items

compare_products

Flipkart

Live details for 2–5 pids/URLs

get_similar_products

Flipkart

Similar / frequently bought together

amazon_autosuggest

Amazon.in

Keyword completions (typo-tolerant)

amazon_search

Amazon.in

ASIN, URL, selling price, MRP, rating

amazon_list_search_filters

Amazon.in

rh refinements (brand, price, RAM, …)

amazon_get_product

Amazon.in

Price, stock, variants, reviews, delivery ETA

amazon_check_delivery

Amazon.in

FREE / fastest delivery at a 6-digit PIN

compare_marketplaces

both

Parallel Flipkart + Amazon search

Typical flow: suggest → search → filter if noisy → get_product / amazon_get_product on the shortlist. Cross-store: compare_marketplaces, then fetch the winning pid and ASIN.

Related MCP server: PriceHunt MCP Server

Install

git clone https://github.com/Nuu-maan/shopping-mcp.git
cd shopping-mcp
uv sync

Run over stdio:

uv run shopping-mcp

Connect an agent

Grok (~/.grok/config.toml or a project .grok/config.toml):

[mcp_servers.shopping]
command = "uv"
args = ["run", "--directory", "/absolute/path/to/shopping-mcp", "shopping-mcp"]
enabled = true
grok mcp add shopping -- uv run --directory /absolute/path/to/shopping-mcp shopping-mcp

Claude Desktop / other stdio MCP hosts:

{
  "mcpServers": {
    "shopping": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/shopping-mcp", "shopping-mcp"]
    }
  }
}

Restart the host after adding the server.

Layout

src/shopping_mcp/
  server.py
  flipkart/
  amazon/

Flipkart talks to 2.rome.api.flipkart.com (/api/4/page/fetch, /api/4/discover/autosuggest). Amazon uses public amazon.in search HTML, /suggestions, /dp/{asin}, and Glow address-change for PIN codes.

No store login is required for the supported tools.

Limits

These are not implemented (no captured request to implement from):

  • Flipkart pincode / delivery date

  • Flipkart full review bodies, Q&A, other sellers

  • Amazon “see all reviews” pagination

  • Amazon other-seller / New & Used listing

Store HTML and private APIs change. Amazon may serve a bot check or captcha; the client retries the interstitial and errors on captcha instead of inventing data.

License

MIT

Available Tools

12 tools
amazon_autosuggestA

Correct and complete an Amazon.in search query.

Use this when the shopper's wording is messy (typos like 'lonovo loq'). Returns keyword suggestions to feed into amazon_search or compare_marketplaces.

Inputs: query (required). Returns: result

amazon_check_deliveryA

Check Amazon.in delivery ETA for an ASIN at a 6-digit pincode.

Uses the Glow address-change call from the product-page HAR, then reloads the product HTML for FREE / fastest delivery times.

Inputs: pincode (required), asin_or_url (required). Returns: result

amazon_get_productA

Fetch one Amazon.in product with live price, specs, delivery, variants, and reviews.

Accepts an ASIN such as B0DW495WY3 or a /dp/ product URL from amazon_search. Optional pincode (6-digit Indian PIN) sets delivery location via Amazon Glow and refreshes ETA (FREE delivery date, fastest slot, cutoff).

Inputs: asin_or_url (required), pincode. Returns: result

amazon_list_search_filtersA

List Amazon.in refinements for a query.

Returns rh strings (p_36 price, p_123 brand, RAM, screen size, …). Pass those as extra_rh on amazon_search. Facets change by category.

Inputs: query (required). Returns: result

autosuggestA

Correct and complete a Flipkart search query.

Use this first when the shopper's wording is messy (typos, missing brand/model). Returns suggested queries, category-scoped searches, and sometimes product matches.

Inputs: query (required). Returns: result

compare_marketplacesA

Search Flipkart and Amazon.in in parallel for the same query.

Use this when the shopper wants the cheaper/better listing across stores. Returns the top organic cards from each marketplace with live selling_price. Follow up with get_product (Flipkart pid) and amazon_get_product (ASIN) on the shortlist before recommending a buy.

Inputs: query (required), limit, max_price, min_price. Returns: result

compare_productsA

Fetch 2–5 Flipkart products in parallel for a side-by-side pick.

Pass pids or product URLs from search_products. Returns the same live fields as get_product for each item. Use this instead of mixing remembered prices with a new search.

Inputs: pid_or_urls (required). Returns: result

get_productA

Fetch one Flipkart product with live price, specs, offers, and reviews.

Accepts a product URL from search results, or a pid such as COMGZW3FSJHF2SEU. Optional listing_id (LST...) selects a specific seller listing.

Returns selling price, MRP, bank-offer price if present, stock, seller, highlights, specification pairs, review titles, similar products, and images. Review bodies and pincode-accurate delivery ETAs are not in this capture — titles/aspect ratings are. Quote the numeric selling_price from this payload.

Inputs: pid_or_url (required), listing_id. Returns: result

get_similar_productsA

Return Flipkart's similar / you-may-also-like products for a pid or url.

Inputs: pid_or_url (required). Returns: result

list_search_filtersA

List Flipkart facets for a query so you can refine search_products.

Returns filter ids (brand, price_range, processor, rating, availability, ...) and each option's params string. Pass those params as extra_facets on search_products. Facets change by category — always call this for the current query instead of guessing facet names.

Inputs: query (required). Returns: result

search_productsA

Search Flipkart and return live product cards.

Each card includes pid, url, selling price, MRP, discount, rating, stock, key specs, and badges (bank offer, only 1 left, exchange, etc).

sort: relevance | popularity | price_asc | price_desc | newest min_rating: 4 means '4★ & above'. brands: e.g. ["ASUS","HP"]. extra_facets: raw params from list_search_filters, like "facets.processor[]=Intel Core i7" or "facets.ssd_capacity[]=512 GB".

Inputs: query (required), page, sort, brands, +5 more. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.2/5.0

    Scored across 12 tools

    Disambiguation5/5

    Each tool has a clear and distinct purpose, with marketplace prefixes (Flipkart vs Amazon) and specific actions (search, get, compare, autosuggest, check delivery). No overlap or ambiguity between tools.

    Naming Consistency4/5

    Most tools follow a verb_noun pattern (e.g., search_products, get_product, list_search_filters). A few tools like 'autosuggest' and 'amazon_search' lack a noun object, but the pattern is still clear and readable. The prefixing by marketplace is consistent.

    Tool Count5/5

    12 tools is well-scoped for a shopping assistant covering two major marketplaces. Each tool serves a distinct step in the shopping workflow (search, refine, compare, check delivery), and none feel redundant or unnecessary.

    Completeness5/5

    The tool surface covers the full shopping journey: query correction, search with filters, product details, cross-marketplace comparison, and delivery estimation. The only missing operations (e.g., cart management) are typically outside the scope of an MCP server and would be handled by the storefront.

    Maintenance

    ActivityStale
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      B
      maintenance
      Enables searching and comparing product prices across major Indian quick-commerce and e-commerce platforms, ranking results by landed price, returning product links, storing price history, and providing 15–30 day price direction forecasts with best-buy recommendations.
      -
    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables AI assistants to discover and search deals from Indian e-commerce stores like Amazon and Flipkart, with tools for trending, latest, category/price/store filtering, deal details, and affiliate link generation.
      -