Forge Geoapify MCP Server
Remote MCP server for business lead discovery via Geoapify, exposing search_businesses and get_business_details tools with deduplication, pagination, and secure API key handling.
README
Forge Geoapify MCP Server
Remote MCP server for Forge Lead Engine discovery.
It exposes exactly two tools:
search_businessesget_business_details
Security model
GEOAPIFY_API_KEYstays only on the server.- Fleet authenticates to the MCP server with
Authorization: Bearer <MCP_AUTH_TOKEN>. - The server never exposes the Geoapify secret in tool outputs, logs, or error bodies.
What it does
- Searches Geoapify Places for candidate businesses.
- Normalizes only fields actually returned by Geoapify.
- Deduplicates by place ID and stable business/location fingerprints.
- Supports pagination, retries, rate-limit handling, timeouts, and malformed-response handling.
- Returns records for downstream Exa verification, Apollo enrichment, qualification, and scoring.
- Keeps outbound email, WhatsApp, follow-ups, and any sales sending disabled.
Important workflow rule
A missing website from Geoapify does not mean the business has no website.
Downstream pipeline should be:
Geoapify discovery → dedupe → Exa website verification → Apollo contact enrichment → qualification → scoring → research → approval queue
No outbound email / WhatsApp behavior is included in this server.
Local run
-
Install dependencies:
npm install -
Copy environment file:
cp .env.example .env -
Set
GEOAPIFY_API_KEYandMCP_AUTH_TOKENin.env. -
Start the server:
npm run dev -
Health check:
curl -s http://localhost:8787/healthz -
MCP endpoint:
curl -s -H "Authorization: Bearer $MCP_AUTH_TOKEN" http://localhost:8787/mcp
Deployment
Recommended easy hosting
Use Render for the simplest free/easy setup.
Steps
- Create a web service from this repository.
- Set environment variables:
GEOAPIFY_API_KEYMCP_AUTH_TOKEN
- Use the Dockerfile or build command:
- Build:
npm install && npm run build - Start:
npm start
- Build:
- Ensure HTTPS is enabled.
- Expose
/mcpand/healthz.
Recommended URL format
- Base:
https://<your-service-domain> - MCP endpoint:
https://<your-service-domain>/mcp - Health:
https://<your-service-domain>/healthz
Fleet custom MCP configuration
Use a Custom MCP server with:
- URL:
https://<your-service-domain>/mcp - Auth mode: Static Headers
- Header:
Authorization: Bearer <MCP_AUTH_TOKEN>
Do not paste GEOAPIFY_API_KEY into Fleet. It stays only in the remote server environment.
Health check
The health endpoint should return:
{ "ok": true }
If the server is behind a platform health probe, point it at /healthz.
Manual MCP test request
After deployment, the exact discovery test request is:
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "search_businesses",
"arguments": {
"country": "UAE",
"city": "Dubai",
"category": "dental clinics",
"max_results": 10
}
}
}
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。