Procurement MCP
Searches AliExpress retail listings in federated marketplace search and reverse image search, returning matching supplier listings with price, MOQ, and supplier info.
Searches Amazon AU listings in federated marketplace search, with optional Keepa price history and relevance filtering, and is included in reverse image search across global retail.
Searches eBay AU listings via the eBay Browse API, including real sold-price data that feeds price history and using seller ratings in trust ranking.
Searches Facebook Marketplace listings via a hosted API, supporting Australian location targeting for local purchases.
Searches Gumtree AU classifieds listings via a hosted scraper API.
Searches Taobao listings in federated marketplace search vai TMAPI, with item search and detail, price normalisation to AUD, and shipping ETA.
Included as a global retail source in reverse image search, matching product photos to Walmart listings with price and supplier data.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Procurement MCP find the cheapest 1L milk at Coles or Woolworths"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Procurement MCP
Procurement MCP is a Model Context Protocol server with 7 tools. search_offers searches up to 16 marketplaces at once and returns one ranked list. Each Offer carries a delivery status and a Landed cost in AUD for the Destination, which is Brisbane 4000 by default.
Every source is a hosted API that the server calls with a key. With no keys set, the ozbargain and cheapies deal feeds work. Every other Marketplace returns a Diagnostic that names the variable to set.
Run it
git clone https://github.com/belikh/procurement
cd procurement
uv sync
uv run python server.pySee also
Available Tools
7 toolsget_offer_detailA
Fetch one offer by id.
taobao, tmall, 1688, jd and pdd ids are fetched live from TMAPI, and ebay_au
ids from the eBay Browse API. Other marketplaces have no detail endpoint:
the offer is returned as it was in the most recent search that contained it.
Inputs: offer_id (required). Returns: result
get_price_historyA
Amazon price history for one product, from Keepa.
Returns current, 30-day and 90-day average, and all-time low and high for
the price sold by Amazon itself. Needs KEEPA_API_KEY and KEEPA_DOMAIN.
Inputs: offer_id_or_asin (required), days, include_series. Returns: result
healthA
Report which marketplaces can run, with no secrets.
Each marketplace has: ready (its key is set), missing (the unset key),
source (the API or actor it calls) and status. Status is verified (a live
run matched the parser), unverified (written from docs), degraded (returns
nothing without extra setup) or unsupported (no source yet).
Inputs: no named parameters. Returns: object
latest_dealsB
List the newest community-posted deals, most upvoted first.
Reads the sites' own RSS feeds, which hold about the latest 100 deals each. Each deal carries votes, comment count, expiry and the price parsed from its title. A feed that fails adds one kind='diagnostic' entry.
Inputs: query, sources, min_votes, max_results. Returns: result
search_by_imageA
Find supplier listings that match a product photo.
Give it a photo, for example from a Gumtree or Facebook listing, and it returns matching listings with price, minimum order and supplier name. Runs a hosted Apify actor that costs about $0.10 per search.
Inputs: image_url (required), destination, max_results. Returns: result
search_historyA
Read past searches from the Postgres search log or cache, newest first.
Needs DATABASE_URL. Without a database the server keeps no history.
Inputs: limit, source, since_hours, include_offers, +1 more. Returns: result
search_offersA
Search marketplaces for a product and return ranked offers.
The result lists buyable offers first (kind='offer', ranked by `sort`), then
past sale prices (kind='comp'), then one kind='diagnostic' entry for every
requested marketplace that returned no offers. A diagnostic's
diagnostic_code is one of: missing_key, timeout, upstream_error, quota,
unsupported, degraded, unsubscribed, no_results.
Each offer has shipping.delivery (direct, pickup, via_forwarder or unknown)
for the destination postcode and landed_aud, the total for `qty` delivered.
landed_basis says how firm that total is: exact, estimated (China offers,
from a forwarder rate and an assumed weight) or item_only (shipping cost not
stated). Offers that cannot reach the destination are left out and counted
in a not_deliverable diagnostic.
All marketplaces run at once under one deadline. Hosted scraper sources
take 30 to 90 s, so a short deadline reports them as timeout.
Inputs: query (required), qty, sort, ship_to, +9 more. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 7 tools
Each tool has a largely distinct purpose: live offer search, past search log, marketplace health, single-offer detail, image-based supplier lookup, community deals, and Amazon price history. There is mild conceptual overlap between search_offers, latest_deals, and get_price_history (all surface prices/deals), but the descriptions distinguish their distinct data sources clearly enough.
Consistent snake_case throughout, with most tools following a verb_noun pattern (search_offers, search_history, get_offer_detail, get_price_history, search_by_image). Two outliers (health, latest_deals) use bare noun/noun-phrase naming, a minor deviation but still readable.
Seven tools is well-scoped for a procurement search server; each covers a distinct capability (search, detail, history, diagnostics, image search, deals, price history) with no filler. Nothing feels redundant or missing at this count.
The surface covers the core search-inspect-history-diagnose lifecycle plus specialized lookups (image, deals, price history). It lacks any write/ordering, saved-search or alert capability, and there's no explicit marketplace-listing or comparison tool, but these are reasonable gaps for a read-oriented search server.
Maintenance
Related MCP Connectors
Routes natural-language shopping queries to merchant storefronts, returns normalized results.
Cross-merchant product search with real price history, comparisons, and demand signals.
Semantic product search over Singapore e-commerce, ranked by an auditable entropy-based value-score
Product search for AI shopping agents: prices and buy links from big retailers. No key needed.
Related MCP Servers
- FlicenseAqualityDmaintenanceEnables AI-powered shopping assistance by analyzing natural language shopping queries and automating product searches on multiple e-commerce platforms.15-
- AlicenseAqualityCmaintenanceEnables querying and comparing prices, availability, ratings, reviews, and seller details from major Russian and Chinese marketplaces (Wildberries, Ozon, Yandex Market, and others) without requiring API keys, via a unified MCP interface.2425MIT
- AlicenseAqualityBmaintenanceUnifies Taobao, JD, and Pinduoduo product search, item detail lookup, and affiliate link conversion (including coupons) into one set of tools exposed to MCP clients, using only official open-platform APIs. Runs in mock mode with realistic fake data until per-platform credentials are supplied, so the full flow can be wired up and tested before appkeys are approved.44MIT
- AlicenseBqualityBmaintenanceEnables AI clients to search, compare, and register products by integrating objective specifications, review-derived experiential attributes, and sales conditions such as price, shipping, and stock.11GPL 3.0