Remote MCP Connection Guide

Access FUDOSAN DB directly from AI agents

What is MCP

MCP (Model Context Protocol) is a standard protocol for AI agents to connect to external data sources. From MCP-capable clients such as Claude.ai, ChatGPT, Claude Desktop, and Cursor, you can query FUDOSAN DB's real-estate data in natural language.

Ask questions like "What's the condo market in Shibuya?" or "Estimate the rent for this unit" — data retrieval, analysis, and rent estimation all run from one prompt.

Setup

Claude.ai and ChatGPT work with Google-account authentication only (no API key required). Other clients need an API key.

Claude.ai

No API key. Just Google authentication.

Connection steps →

ChatGPT

No API key. Just Google authentication.

Connection steps →

Claude Desktop / Claude Code (API key method)

Get an API key

Create an account on the API key page to issue a key. Free plan is available.

Install npx mcp-remote

Requires Node.js 18+. npx auto-installs it — no prior install needed.

Edit the Claude Desktop config

Add the following to claude_desktop_config.json.

// claude_desktop_config.json
{
  "mcpServers": {
    "fudosandb": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://fudosandb.jp/mcp",
        "--header",
        "X-API-Key: YOUR_API_KEY"
      ]
    }
  }
}
        

Restart Claude Desktop

Save the config and restart Claude Desktop to activate the MCP connection. FUDOSAN DB will appear in the tool list.

Available tools

search_areas

Search municipalities by prefecture / city name. Returns area profiles (transaction count, median, land price, etc.).

get_area_profile

Get detailed area profile by municipality code.

get_price_trends

Transaction price trends for a given area. Filterable by property type.

get_land_price_trends

Public land price trends for a given area (annual).

get_rankings

Rankings by metric: transaction count, median price, land price, etc.

list_municipalities

List all municipalities. Filterable by prefecture.

estimate_rent

AI rent estimation from area, size, age, layout, etc. Accuracy (MAPE), measured under this tool's own input conditions on properties unseen during training (measured 2026-09-05, v5.3-areafix, duplicate rows excluded): approx. 10.7% (9 urban prefectures) / approx. 10.1% (38 rural prefectures) with no optional fields supplied, improving to approx. 9.4% (urban) / 9.5% (rural) with station_name alone, and approx. 9.2% when both station_name and category are supplied. Estimates for 88 m² and above are outside the range covered by the accuracy figures above because that band has few training listings (the response carries a warning and confidence.level="low"; treat as reference values). Accuracy figures for the 88 m²+ band are being re-measured after the September 2026 retraining.

simulate_yield

Real-estate investment cash-flow simulation. Yield / CF / IRR for up to 10 years.

predict_rent_change

Forecast municipality-level rent change (1y / 3y). 22 features across macro indicators, future population, and regional characteristics. MAE 1.6pp, direction accuracy 83%.

predict_future_rent

Future rent forecast. Phase 2 rent estimate × Phase 3 change rate. Starting point for medium-term yield analysis.

J-REIT tools

search_reit_properties

Cross-REIT property search. Filter by area, use type, cap-rate range. 6,900+ properties / 73 REITs.

get_reit_property_history

Per-property period history (occupancy / rent / appraisal / cap rate), up to 10 periods.

get_reit_portfolios

Per-REIT portfolio summary (AUM, avg cap rate, occupancy).

search_reit_sales

J-REIT disposition events: sale price, book value, gain/loss, price per tsubo, sponsor-transaction flag.

search_reit_acquisitions NEW

J-REIT acquisition events (unified transaction-case format). Acquisition date / price / price per tsubo / seller / appraisal / NOI cap / sponsor-transaction / holding years.

get_hotel_performance

Hotel-property monthly operations (ADR / RevPAR / GOP / occupancy, JPY and USD).

get_reit_appraisals

Periodic appraisal details (direct-cap price / DCF / discount rate / terminal cap rate / appraisal NOI).

get_reit_appraiser_bias NEW

Cap rate evaluation bias by appraiser. Deviation from peer median aggregated by use_type × region × year. Positive = conservative, negative = aggressive.

get_reit_appraisal_impairment_alerts NEW

Early-warning alerts on appraisal declines. Period-over-period drops of -5%+ or cap rate spikes of +0.3pp classified as severe / alert / watch. For activist / institutional investor screening.

get_reit_demographic_risk NEW

Per-REIT future population risk aggregated from 250m-mesh XKT013 (2020-2050) × geocoded REIT property addresses. Includes acquisition-price-weighted weighted_pop_change_2050, pct_high_plus, and pct_aum_high_risk to quantify demographic decline exposure.

search_reit_tenants

Top-tenant search per property: tenant name, industry, leased area, contract form. Anonymized tenants included.

get_reit_sponsors NEW

J-REIT × sponsor-company master. Reference data for sponsor-transaction detection.

search_reit_counterparties NEW

Normalized master of sellers / buyers / SPCs. Useful for deduplication and identifying major buyers/sellers.

search_corporate_real_estate_transactions NEW

Non-REIT corporate real-estate transaction cases. LLM-structured extraction from EDINET annual-report facility plans and sale notes. Filterable by transaction type, amount, counterparty.

get_reit_sponsor_performance NEW

Sponsor consolidated financials × REIT disposition, by fiscal year. Useful for "properties concentrated during a sponsor-decline year" analysis.

get_sponsor_pipeline_forecast ML

Calibrated probability score (0-100) for sponsor → REIT transfers next fiscal year. Calibrated LR v1 / Leave-One-Sponsor-Out AUC=0.75. Per-feature contribution (log-odds) and confidence label (high/medium/low) included.

search_corporate_real_estate_holdings NEW

Listed-company real-estate holdings × land price × 2050 population forecast × hazard risk. 267K facilities across 4,322 companies.

search_reit_hazard_risk NEW

Hazard-zone aggregation in the municipality where each J-REIT property sits (steep slope / landslide / tsunami / flood / storm surge). Values are municipality-level counts, not a parcel-level overlay, and do not indicate whether the property's parcel intersects a zone (properties in the same municipality share the same values).

Prompt catalog

Prompts you can drop straight into an MCP-connected Claude. Copy and run as-is.

Market research

Show the condo transaction price trend in Shibuya-ku. Plot the quarterly median for the past 5 years.

get_price_trends → filter property_type: condo, render as time series

Compare the condo market in Setagaya, Suginami, and Meguro in a single table.

Pull get_area_profile for each and assemble a comparison table

Rank the wards of Yokohama by land price — residential and commercial separately.

search_areas for Yokohama wards, then sort by land price

Investment analysis

Top 10 Tokyo 23 wards by land-price YoY growth for the most recent year.

get_rankings → land_price_growth

Plot condo prices and land prices for Chuo-ku, Osaka side-by-side and interpret from an investment-timing angle.

Combine get_price_trends + get_land_price_trends

Comparative market analysis of Chiyoda, Minato, and Chuo wards, written as a short report.

Aggregate multiple tools into a combined report

J-REIT acquisition case analysis

List logistics J-REIT properties acquired from sponsors with acquisition-time NOI cap ≥ 4.5%. Include price per tsubo.

search_reit_acquisitions → use_type=logistics + is_sponsor_transaction=true + cap_rate_min=4.5

Top 20 central-Tokyo office properties acquired since 2024 where the seller was NOT the sponsor (arm's-length).

search_reit_acquisitions → use_type=office + is_sponsor_transaction=false + acquisition date filter

For properties sold within 5 years of acquisition, compute average holding period and acquisition-price → sale-price delta (unrealized gain proxy) by use type.

search_reit_acquisitions (is_sold=true) → aggregate disposition P&L

Sponsor × performance (activist lens)

Line up Mitsui Fudosan's financial performance year-by-year with the properties transferred to Mitsui-sponsored REITs (e.g., Nippon Building Fund). I want to see if transfers cluster in decline years.

get_reit_sponsor_performance → sponsor_name=Mitsui Fudosan + only_sponsor_to_reit=true

List sponsor-to-REIT disposition cases where sponsor performance YoY declined (sponsor_in_decline=true), sorted by amount.

get_reit_sponsor_performance → only_decline=true

For the top 10 REIT sponsors, fetch 5-year ROE trajectories and check correlation with REIT-disposition amounts.

get_reit_sponsors (master) → get_reit_sponsor_performance (join)

List the top 10 sponsors most likely to transfer properties to their REITs next fiscal year, with score and reasoning.

get_sponsor_pipeline_forecast → confidence=high. Decompose via feature_contributions_logodds.

What is Mitsui Fudosan's predicted score for next fiscal year? Show the per-feature contribution.

get_sponsor_pipeline_forecast → sponsor_name=Mitsui Fudosan

Counterparty master (buyers / sellers / SPCs)

Top 20 entities that have sold ≥ 10 J-REIT properties over the past 5 years. Split sponsors vs. non-sponsors.

search_reit_counterparties → role=seller + min_transactions=10

Count how often SPC-style names (e.g., "XX Tokutei Mokuteki Kaisha") appear as seller/buyer. I want to surface the major bridge-fund operators.

search_reit_counterparties → filter SPC-name patterns

Timeline of transactions where Sumitomo Mitsui Trust Bank shows up as trustee/buyer — only rows with a resolved corporate number.

search_reit_counterparties → name=Sumitomo Mitsui Trust + has_corporate_number=true

Disaster risk × REIT portfolio

Extract hotel properties held by J-REITs located in municipalities with tsunami inundation zones. Include operating REIT and acquisition price.

search_reit_hazard_risk → has_tsunami=true + use_type=hotel

Rank REITs by total hazard-zone count to check whether disaster risk is diversified.

search_reit_hazard_risk aggregated by REIT

Properties with ≥ 3 overlapping flood + landslide zones, broken down by use type. I want to see if logistics concentrates the risk.

search_reit_hazard_risk → min_total_zones=3 + has_flood + has_landslide

Listed-company real-estate holdings

List all facilities owned by Toyota Motor nationwide, sorted by surrounding commercial land price. Attach 2050 population forecast and flood risk.

search_corporate_real_estate_holdings → company_name=Toyota Motor

In Tokyo 23 wards, show listed-company facilities with land book value > ¥10B and surrounding commercial land price ≥ ¥1M/m². Looking for unrealized-gain candidates.

search_corporate_real_estate_holdings → prefecture=Tokyo + min_land_book_value=10000 + min_commercial_price=1000000

Top 20 listed companies by number of facilities located in areas with 2050 population decline ≥ 30%. Useful for future-idle-asset analysis.

search_corporate_real_estate_holdings → max_population_change=-30

Non-REIT corporate transactions (LLM-extracted from annual reports)

List non-REIT listed-company real-estate dispositions ≥ ¥10B over the past 2 years, along with the counterparty.

search_corporate_real_estate_transactions → transaction_type=disposed + amount_min=10000

Aggregate announced large real-estate acquisitions still in "planned" status (new plants, etc.) by industry.

search_corporate_real_estate_transactions → transaction_type=planned_acquisition

For logistics facilities sold by non-REIT corporates, show the buyer (counterparty). I want to see if it's mostly J-REIT / fund vs. third-party.

search_corporate_real_estate_transactions → filter logistics via property_name / location + counterparty_name analysis

Cross-DB applied examples

Properties transferred in sponsor-decline years (sponsor_in_decline=true) AND located in ≥ 3 hazard zones. I want to surface the "forced-disposition" pattern.

Join get_reit_sponsor_performance with search_reit_hazard_risk

For listed companies that just announced a large real-estate sale, pull land price and population dynamics for their other facilities. I'm reading continuity in divestment strategy.

search_corporate_real_estate_transactions + search_corporate_real_estate_holdings 2-hop

Logistics J-REIT acquisitions located in flood-risk municipalities only. Line up acquisition-time NOI cap. I want to verify the risk premium is getting priced.

search_reit_acquisitions (use_type=logistics) + search_reit_hazard_risk (has_flood=true)

Using from Claude Code

The same MCP connection works from Claude Code. Add the config to ~/.claude/settings.json.

// ~/.claude/settings.json
{
  "mcpServers": {
    "fudosandb": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://fudosandb.jp/mcp",
        "--header",
        "X-API-Key: YOUR_API_KEY"
      ]
    }
  }
}
        