NDI-MCP-Server
Northeast Deal Intel — MCP 服务器
将任何支持 MCP 的 LLM(Claude、Cursor、Continue)连接到东北部地区 14,000 多条经过 AI 评分的商业地产交易和 100,000 多条已成交案例。
你可以做什么
向你的 LLM 询问自然语言问题,例如:
“查找康涅狄格州 (CT) 资本化率超过 7%、价格低于 200 万美元的工业地产”
“我有一个 140 万美元的 1031 置换项目,45 天内到期。有哪些 NNN 零售地产适合?”
“将这个哈特福德仓库的资本化率与近期成交案例进行比较”
“为这笔交易评分:215 Main St, Windsor CT, 320 万美元工业地产, 7.8% 资本化率”
“目前康涅狄格州哪些交易的出售概率最高?”
Related MCP server: nyc-property-intel
公开工具
工具 | 描述 | 所需等级 |
| 按州、类型、评分、价格、资本化率查找活跃房源 | 任意 |
| 完整的交易详情 + 评分细分 + 出售信号 | 任意 |
| 100,000 多条已成交交易,用于基准比较 | agent_starter |
| 提交任何交易进行 AI 评分 | agent_pro |
| 按州/类型划分的资本化率 + 每平方英尺价格 (PSF) 基准 | 任意 |
| 针对置换交易的筛选搜索 | 任意 |
| 特定房源的出售概率 | agent_starter |
| 州级市场概览 | 任意 |
设置
1. 获取 API 密钥
在 northeastdealintel.com/agent-api.html 注册
Agent Starter ($49/月) —
search_deals,get_deal,search_comps,get_sell_signal,get_market_summaryAgent Pro ($149/月) — 以上所有功能 +
score_deal(提交任何交易进行 AI 评分)Agent Enterprise ($499/月) — 完全访问权限 + 批量导出 + 自定义成交报告
2. 安装依赖
pip install mcp httpx3. 配置 Claude Desktop
添加到 ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) 或
%APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"northeast-deal-intel": {
"command": "python3",
"args": ["/path/to/ndi_mcp_server.py"],
"env": {
"NDI_API_KEY": "your_api_key_here"
}
}
}
}重启 Claude Desktop。你将在工具选择器中看到 NDI 工具。
4. 为其他 MCP 客户端进行配置
该服务器使用 stdio 传输(标准)。任何支持 MCP 的客户端工作方式相同——只需将其指向 ndi_mcp_server.py 并在环境变量中设置 NDI_API_KEY。
Cursor: 添加到项目根目录下的 .cursor/mcp.json。
Continue: 添加到 ~/.continue/config.json 中的 mcpServers 下。
对话示例
1031 置换
“我以 180 万美元卖掉了一个康涅狄格州的商业中心,需要在 45 天内找到置换房产。帮我找资本化率高于 7%、价格在 150 万到 220 万美元之间的 NNN 零售地产。”
市场研究
“目前哈特福德与费尔菲尔德县的工业地产平均资本化率分别是多少?请展示每个子市场中排名前 5 的交易。”
交易承销
“为这笔交易评分:45 Industrial Dr, Wallingford CT。要价 410 万美元,47,000 平方英尺仓库,单租户 NNN,6.8% 资本化率,租约剩余 4 年。”
寻找困境资产
“帮我找康涅狄格州出售概率高的交易——我在寻找急于出售的卖家。重点关注工业和多户住宅。”
API 基础 URL
https://api.northeastdealintel.com
使用环境变量覆盖:NDI_API_BASE=https://api.northeastdealintel.com
问题
Available Tools
8 toolsfind_1031_candidatesA
Find deals suitable for a 1031 exchange. Filters for income-producing properties with clean structures, appropriate price bands, and NNN/NN lease profiles. Pass the exchanger's target price range and timeline for best results.
Inputs: state, max_price, min_price, min_cap_rate, +2 more.
get_dealA
Get full details for a specific deal by ID, including complete AI scoring breakdown, green/red flags, sell probability signal, distress tier, and 1031 suitability.
Inputs: deal_id (required).
get_market_benchmarksA
Get cap rate and price/SF benchmarks for a state and property type, derived from closed comps. Use to determine if a deal is priced above or below market.
Inputs: state (required), property_type.
get_market_summaryB
Get a market summary for a state: total active listings, score distribution, average cap rate, deal count by property type, and top submarkets.
Inputs: state (required).
get_sell_signalA
Get the sell probability signal for a listed property — the likelihood it transacts in the next 6 months based on days-on-market, ownership age, distress tier, and score. High sell signal = motivated seller, potential to negotiate. Requires agent_starter tier.
Inputs: listing_id (required).
score_dealA
Submit a deal for AI scoring. Returns a 1-10 score, green flags, red flags, market benchmarks, and investment thesis. Use when evaluating a deal not yet in the NDI database. Requires agent_pro tier.
Inputs: state (required), property_type (required), asking_price (required), address, +4 more.
search_compsA
Search 100,000+ closed commercial transactions for comp data. Use to benchmark a deal's price/SF or cap rate against actual recent sales. Requires agent_starter tier.
Inputs: state (required), property_type, min_price, max_price, +5 more.
search_dealsA
Search active commercial real estate listings across the Northeast. Returns AI-scored deals with cap rates, pricing, green/red flags, and sell signals. Use this to find deals matching an investor's criteria, scout a submarket, or identify opportunities for a 1031 exchange.
Inputs: state, property_type, min_score, max_score, +5 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 8 tools
Every tool has a clearly distinct purpose with no ambiguity. For example, find_1031_candidates focuses on filtering for 1031 exchange suitability, while search_deals is a broader search for active listings, and get_deal retrieves detailed information for a specific deal. The descriptions reinforce these unique roles, ensuring agents can easily differentiate between them.
All tool names follow a consistent verb_noun pattern using snake_case, such as find_1031_candidates, get_deal, and search_comps. This uniformity makes the toolset predictable and easy to navigate, with no deviations in naming conventions across the eight tools.
With 8 tools, the count is well-scoped for a commercial real estate analytics server. Each tool serves a specific function, from searching and scoring deals to retrieving market data and benchmarks, ensuring comprehensive coverage without unnecessary bloat or gaps.
The toolset provides strong coverage for commercial real estate analysis, including deal discovery, scoring, benchmarking, and market insights. A minor gap exists in update or management operations for deals, but the core workflows for evaluation and research are fully supported, allowing agents to perform essential tasks effectively.
Related MCP Connectors
MCP server giving Claude AI access to 22+ NYC public-record databases for real estate due diligence
Kliv is an AI web-app builder. It hosts each app with its own database, edge functions, commerce and email. Connect Claude Code or Cursor to Kliv's MCP server, sign in with your Kliv account, and your AI tool can work on your apps with the same tools the Kliv builder uses. It can edit an app's files, change the database, write and run edge functions, set up products and subscriptions, and run your tests. Then it can build and publish the app. Other MCP clients connect the same way. Your AI tool runs on your own AI subscription, so this work doesn't use your Kliv usage. Kliv still hosts the app and everything behind it. Each person signs in as themselves and reaches only the apps they can edit. The account owner sees every connected tool and can remove one at any time. The address is https://mcp.kliv.dev/mcp for everyone. Sign-in is OAuth in your browser, with your Kliv account. There are no API keys. Setup guide: https://docs.kliv.dev/integrations/mcp
Hosted Klaviyo MCP server for Claude, ChatGPT and Cursor, with each action switched per account.
Marketo MCP server for AI. 130 tools to operate Marketo from Claude, Cursor, or ChatGPT.
Related MCP Servers
- AlicenseAqualityDmaintenanceOpen-source MCP server providing real estate regulatory intelligence (zoning, permits, entitlements, deal scoring) for US properties, enabling AI agents to access 10 callable tools.125 npmMIT
- AlicenseAqualityBmaintenanceMCP server for NYC real estate due diligence. Lets Claude query 22+ NYC public-record databases — DOB/HPD/ECB violations, ACRIS deeds, DOF sales, 311 complaints, FDNY incidents, NYPD complaints, marshal evictions, PLUTO, rent stabilization — in plain English.187MIT
- AlicenseNot gradedqualityDmaintenanceProvides live commercial real estate data (rates, demographics) and analysis tools (DCF, rent roll parsing, lease abstraction, IC memo generation) within Claude Desktop.1MIT
- FlicenseNot gradedqualityDmaintenanceA production-grade MCP server enabling Claude to perform comprehensive NJ real estate workflows including property search, valuation, neighborhood intelligence, investment analysis, and agent tools via 20 tools and 15+ data sources.-