search_dealsLABSCO SUMMARY
Deal screening as a question rather than a spreadsheet.
The listings are already scored, so a buy box stated in a sentence comes back ranked instead of as a list to work through — and the comps and benchmarks needed to sanity-check the answer are in the same server.
What it is
A Python MCP server you run locally, which calls Northeast Deal Intel's API for its commercial real estate dataset: active listings with AI scores, closed comparable transactions, market benchmarks, submarket profiles, and 1031 and Opportunity Zone matching. Calls are covered by an API key subscription or paid one at a time over x402.
What you get
- Active listings filtered by state, property type, score, price and cap rate
- One deal in full, with its scoring breakdown and sell signal
- Closed comps for benchmarking, up to 10 per call
- Any deal scored on demand, singly or ten in one batch
- Cap rate and price-per-square-foot benchmarks by state and property type
- Submarket profiles: cap rates, comp velocity and score distribution
- Ranked 1031 replacement candidates, and Opportunity Zone deals flagged for OZ and 1031 crossover
- A state-level market overview
- Current pricing, tier information and x402 setup, free of charge
Requirements
Either an API key subscription — Agent Starter, Agent Pro or Enterprise, each with its own monthly price and daily request allowance — or x402 pay-per-call in USDC on Base mainnet with no signup at all. Per-call pricing is lowest for a deal or benchmark lookup and highest for a scoring run. Coverage is the Northeast US.
Setup effort
Build from source — git clone https://github.com/CREIntel/ndi-mcp-server.git
USE CASES
01
Screening deals against your buy box
State, type, cap rate and price ceiling go into one search; the full deal and its score come back from the next call.
02
Sanity-checking a cap rate you were quoted
Closed comps and state-level benchmarks put the number in context before you negotiate on it.
03
Placing a 1031 exchange before the clock runs out
Ranked replacement candidates for the money you have, with Opportunity Zone crossover flagged.
04
Ranking a shortlist you brought yourself
Submit a deal for scoring, or up to ten at once, and rank them by what comes back.
05
Deciding whether to sell
The sell signal for a listing, read next to the deal's own scoring breakdown.
06
Getting a read on a submarket
Cap rates, comp velocity and score distribution for the submarket, plus the state-level picture around it.
13 tools
Listed in project docs. Your client may show a different list.
get_dealsearch_compsscore_dealget_market_benchmarksfind_1031_candidatesget_sell_signalget_market_summaryget_compsget_submarket_intelget_oz_dealsbatch_score_dealsget_accessBEFORE YOU INSTALL
What to know first
Tools you get
13 tools named in the project's documentation.
Requires
NDI_API_KEY in your environment
Where it runs
All 3 documented clients run the server on your machine.
INSTALLATION
Connect your client
You clone this one yourself.
Server setup
Clone the repository, which holds ndi_mcp_server.py, and install its dependencies
TERMINAL
git clone https://github.com/CREIntel/ndi-mcp-server.git
cd ndi-mcp-server
pip install mcp httpxMCP Clients
Claude Desktop, Cursor, Continue
JSON
{
"mcpServers": {
"northeast-deal-intel": {
"command": "python3",
"args": [
"<YOUR_PATH_TO_NDI_MCP_SERVER>"
],
"env": {
"NDI_API_KEY": "<YOUR_NDI_API_KEY>"
}
}
}
}
