Project
Zoner MCP
A Model Context Protocol server that powers zoning lookups for AI agents. Transform street addresses into zoning data for Lebanon, NH, and plug into any OpenAI Agent Builder or ChatKit workflow.
What It Does
ZonerMCP is a minimal, reusable MCP server template designed for use with OpenAI Agent Builder and other MCP-compatible AI platforms. It uses Streamable HTTP transport for broad compatibility.
The server takes a street address and returns zoning information, enabling AI agents to answer questions like "What zone is 123 Main Street in?" or "Can I build a duplex at this address?"
Design Philosophy
This template is intentionally minimal and UI-agnostic. It provides only the MCP server infrastructure, allowing you to:
- Connect it to any AI interface (OpenAI Agent Builder, Claude, custom UIs)
- Build your own frontend or use existing AI chat interfaces
- Focus on your business logic without UI constraints
- Deploy as a standalone microservice
Tech Stack
Key Features
Streamable HTTP Transport
Works with OpenAI Agent Builder and MCP Inspector
Stateless Client Support
Auto-initializes sessions for clients that don't handle the MCP handshake
CORS Enabled
Ready for cross-origin requests out of the box
Health Check Endpoint
Built-in monitoring and load balancer support
API Endpoints
| Endpoint | Method | Description |
|---|---|---|
| /mcp | POST | Main MCP endpoint (initialize, tools/list, tools/call) |
| /mcp | GET | Server info or SSE stream (with session) |
| /mcp | DELETE | Session termination |
| /health | GET | Health check |
License
ISC License - Open source and free to use, modify, and distribute.