Japan Market MCP — Cross-Shop Price Comparison for AI Agents

An MCP server that lets AI agents (Claude, Cursor, VS Code Copilot, ChatGPT) compare used-good prices across leading Japanese shops in a single tool call. Instead of scraping site by site, an agent queries one server and gets inter-store price gaps for the same product — perfect for reseller arbitrage, price monitoring, and market research on the Japanese second-hand market.

How it works

The server wraps market-scraping and official-API actors (Rakuten Ichiba included) behind one MCP endpoint. Each tool returns price, brand, shop, condition and a link per item, with the shop-pair crossed so you see the price spread at a glance.

Available tools

Tool	Market	Shop pair
`search_camera_market`	Used cameras & lenses	Kitamura + Fujiya Camera
`search_watch_market`	Luxury watches	Jackroad + Kitamura used watches
`search_luxury_market`	Used luxury brands	Komehyo + Jackroad
`search_instrument_market`	Used instruments	Digimart + Ishibashi U-BOX
`search_offmall_market`	General used goods	OffMall (Hard Off official, 800+ stores)
`search_kakaku_prices`	New price comparison	Kakaku.com (aggregated, thousands of shops)
`search_car_market`	Used cars	goo-net (nationwide, by body type)
`search_car_price_stats`	Used car price stats (min/max/avg/median JPY)	goo-net (by model keyword, e.g. N-BOX, Alphard)
`search_rakuten_items`	New retail products (official API)	Rakuten Ichiba 楽天市場 (keyword, price range, sort)
`get_rakuten_ranking`	Best-seller ranking (official API)	Rakuten Ichiba 楽天市場 (overall or by genre)
`get_maff_market_report`	Fresh produce wholesale market report (official MAFF open data)	Japan central markets (旬別, JPY/100kg, YoY%)
`search_maff_price_trend`	Fresh produce wholesale price trend (official MAFF open data)	Multi-market, multi-period comparison
`get_maff_top_movers`	Fresh produce YoY price movers (official MAFF open data)	Japan central markets (supply-shock signal)
`get_maff_markets`	List available MAFF wholesale markets	no charge

Each tool takes a keyword (e.g. SONY α7, ROLEX, Hermes, Fender) and returns results from the crossed shops. Leaving the keyword empty scans the full category.

Pricing

Pay per event — $0.001 per search + actor start fee. Runs finish in seconds, so a full keyword lookup typically costs well under $0.01.

Connect as an MCP server

Run it from the Apify Store (Run button), then point your MCP client at the endpoint. It stays in standby mode.

MCP client config (Claude Desktop / Cursor / VS Code):

config.json

{
  "mcpServers": {
    "japan-market-mcp": {
      "url": "https://fruitful-quintessence--japan-market-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

URL: <your-username>--japan-market-mcp.apify.actor/mcp
Auth: Apify API token in the Authorization: Bearer header (Apify Console → Settings → Integrations)

URL note: use a hyphen in the username (fruitful-quintessence), not an underscore — an underscore hostname will not resolve.

Discovery & registry

Agent install guide: see llms-install.md — grounded guidance for AI agents (Cline, Claude Code, Cursor) that set up this server for a user.
MCP discovery manifest: .well-known/mcp.json advertises the endpoint, auth, tool list, and pricing in the standard MCP Directory discovery format.
Cline MCP Marketplace: submitted for review at cline/mcp-marketplace — once approved it becomes one-click installable from Cline's marketplace.
Apify Store: deployed as the public actor fruitful_quintessence/japan-market-mcp (pay-per-event).

Related Apify Actors

This MCP server connects to 86+ Apify Actors in the Kensho ecosystem for real-time data. Key related actors include:

MCP Connection Examples

6. External run example (fallback)

bash

# If GitHub API returns 403, use curl + git clone
curl -s https://raw.githubusercontent.com/atushi1841/japan-market-mcp/main/README.md > README.patch
# edit README.patch as needed
git clone https://github.com/atushi1841/japan-market-mcp.git
cd japan-market-mcp
cp ../README.patch README.md
git add README.md
git commit -m "update README via external run"
git push

1. Standard I/O (stdio) — Claude Desktop, Cursor, etc.

In your claude_desktop_config.json:

config.json

{
  "mcpServers": {
    "japan-market-mcp": {
      "command": "python",
      "args": ["-m", "src.main"]
    }
  }
}

2. HTTP Server Mode

bash

# Install dependencies
pip install -r requirements.txt

# Run as HTTP server on port 3000
APIFY_TOKEN=your_token python -m src.main --http --port 3000

# Test the endpoint
curl http://localhost:3000/mcp/tools | jq .

3. Direct Python Import

server.ts

from src.main import mcp
# FastMCP instance can be embedded directly

4. n8n / Apify Workflow (via HTTP)

Use Apify's Webhook node to POST JSON-RPC to POST /mcp endpoint.

5. Docker Deployment

Terminal

docker build -t japan-market-mcp .
docker run -p 3000:3000 -e APIFY_TOKEN=your_token japan-market-mcp

Code

"Compare used prices for Sony a7 III between Kitamura and Fujiya"
→ search_camera_market(keyword="SONY α7 III")

"What's the inter-store price gap for a Rolex Submariner?"
→ search_watch_market(keyword="ROLEX Submariner")

"What's the market price for a Hermes Birkin 25?"
→ search_luxury_market(keyword="Hermes Birkin 25")

"What's a used Fender Stratocaster going for?"
→ search_instrument_market(keyword="Fender Stratocaster")

"Show this 旬's wholesale market report for Tokyo (Toyosu) fresh vegetables"
→ get_maff_market_report(market="toyosu")

"Which fresh produce items have spiked in price vs last year at Osaka?"
→ get_maff_top_movers(market="oosakaho")

"Track the price of daikon (だいこん) across Tokyo and Osaka this month"
→ search_maff_price_trend(item="だいこん", markets=["toyosu","oosakaho"])

Use cases

Reseller arbitrage — spot the same product priced differently across shops
Market research — track used prices for cameras, watches, luxury goods, instruments, cars
Export sourcing — goo-net JDM inventory and goo-net price checks for exporters
Collector price checks — condition-ranked physical used items from OffMall

Data sources

Each actor collects factual data only (product name, price, brand, condition, inventory, URL). No photos or descriptions are harvested.
All target sites allow crawling via their robots.txt.
MAFF 青果物卸売市場調査（旬別・市場別） is served from the official 農林水産省 open-data CSV — no scraping, direct download. Published under the Government Standard Terms of Use v2.0 (政府標準利用規約2.0): commercial use allowed with attribution. Price is JPY per 100kg. Data updates are published by MAFF each 旬 (10-day period).

Local development

Terminal

pip install -r requirements.txt
APIFY_TOKEN=apify_api_xxx python -m src.main
# Streamable HTTP endpoint at http://localhost:3000/mcp

License

MIT

Install via Smithery

Connect this MCP server to your AI client (Claude Desktop, Cursor, VS Code) in one command:

Terminal

npx @smithery/cli install atushi1841/japan-market-mcp --client claude

Replace claude with cursor, vscode, or cline for other clients.

Alternatively, install directly from the Smithery registry.

Note: Smithery server listing is pending verification. Once verified, this server will appear in search results with useCount tracking.

Japan Market MCP — Cross-Shop Price Comparison for AI Agents

How it works

Available tools

Pricing

Connect as an MCP server

Discovery & registry

Related Apify Actors

Marketplace & Price Comparison

Retail & E-commerce

Used Goods & Auctions

Local Business & Services

Government & Public Data

Specialized Collections

MCP Connection Examples

6. External run example (fallback)

1. Standard I/O (stdio) — Claude Desktop, Cursor, etc.

2. HTTP Server Mode

3. Direct Python Import

4. n8n / Apify Workflow (via HTTP)

5. Docker Deployment

Use cases

Data sources

Local development

License

Install via Smithery