example-mcp-server-sse
example-mcp-server-sse
README
Calculator Learning Demo - SSE (Legacy) Transport
<div align="center">
</div>
<p align="center"> <strong>⚠️ DEPRECATED TRANSPORT ⚠️</strong><br/> This repository demonstrates the legacy HTTP + SSE transport for <strong>educational purposes only</strong>.<br/> For new projects, please use the modern <a href="https://github.com/yigitkonur/mcp-server-examples/tree/main/streamable-http">Streamable HTTP transport</a>. </p>
🎯 Overview
This repository provides a reference implementation of an MCP server using the classic two-endpoint HTTP + Server-Sent Events (SSE) transport. It is intentionally designed to teach the concepts, complexities, and limitations of this deprecated pattern compared to modern, single-endpoint transports.
🔧 Educational Echo Tool
This server includes an optional educational tool for learning MCP concepts:
- Tool Name: Configurable via
SAMPLE_TOOL_NAMEenvironment variable - Functionality: Simple echo tool that demonstrates basic MCP tool patterns
- Usage: Set
SAMPLE_TOOL_NAME=your_tool_nameto add the tool to the server - Purpose: Provides a minimal example for understanding MCP tool registration and execution
Key Characteristics
- Asymmetric Channels: Utilizes
GET /ssefor a persistent server-to-client event stream and a separatePOST /messagesendpoint for client-to-server commands. - Ephemeral Session State: Each connection establishes a new, isolated session on the server. All state (e.g., calculation history) is stored in memory and is lost when the connection closes.
- No Resumability: If the SSE connection is lost, the session cannot be recovered. The client must establish a new session.
- Network Dependency: Requires careful handling of network proxies and firewalls that may buffer or block SSE streams.
📊 Transport Comparison
This table compares the four primary MCP transport mechanisms demonstrated in the learning series. The implementation in this repository is highlighted.
| Dimension | STDIO | SSE (Legacy) | Streamable HTTP (Stateful) | Streamable HTTP (Stateless) |
|---|---|---|---|---|
| Transport Layer | Local Pipes (stdin/stdout) |
✅ 2 × HTTP endpoints (GET+POST) |
Single HTTP endpoint /mcp |
Single HTTP endpoint /mcp |
| Bidirectional Stream | ✅ Yes (full duplex) | ⚠️ Server→Client only | ✅ Yes (server push + client stream) | ✅ Yes (within each request) |
| State Management | Ephemeral (Process Memory) | ✅ Ephemeral (Session Memory) | Persistent (Session State) | ❌ None (Stateless) |
| Resumability | ❌ None | ❌ None | ✅ Yes (Last-Event-Id) |
❌ None (by design) |
| Scalability | ⚠️ Single Process | ✅ Multi-Client | ✅ Horizontal (Sticky Sessions) | ♾️ Infinite (Serverless) |
| Security | 🔒 Process Isolation | 🌐 Network Exposed | 🌐 Network Exposed | 🌐 Network Exposed |
| Ideal Use Case | CLI Tools, IDE Plugins | ✅ Legacy Web Apps | Enterprise APIs, Workflows | Serverless, Edge Functions |
📐 Architecture and Flow
The legacy SSE transport pattern requires a two-step communication flow. The client first establishes a long-lived GET request to the /sse endpoint to listen for events. The server responds with a unique sessionId, which the client must then include as a query parameter in all subsequent POST requests to the /messages endpoint. This allows the server to route incoming commands to the correct session and event stream.
sequenceDiagram
participant Client
participant Server
Note over Client,Server: Connection Establishment
Client->>Server: GET /sse
Server-->>Client: Establishes text/event-stream connection
Server-->>Client: event: endpoint\ndata: /messages?sessionId=abc-123
Note over Client,Server: Client-to-Server Request
Client->>Server: POST /messages?sessionId=abc-123
Note right of Client: Body: {"jsonrpc":"2.0", "method":"tools/call", ...}
Server-->>Client: 202 Accepted (Response will be sent via SSE stream)
Note over Client,Server: Server-to-Client Response / Notification
Server-->>Client: event: message\ndata: {"jsonrpc":"2.0", "id":1, "result":...}
✨ Feature Compliance
This server implements a limited subset of the MCP Latest Standard to demonstrate the core SSE pattern. Features requiring more complex state or interaction models are stubbed or not implemented.
| Name | Status | Implementation |
|---|---|---|
calculate |
Core ✅ | Basic arithmetic operations (add, subtract, multiply, divide, power, sqrt). |
batch_calculate |
Not Implemented | Returns JSON-RPC error -32601 Method not found. |
advanced_calculate |
Not Implemented | Returns JSON-RPC error -32601 Method not found. |
demo_progress |
Extended ✅ | Simulates progress by logging to the console; designed to show how events would be pushed. |
explain-calculation |
Core ✅ | Returns a Markdown explanation prompt. |
generate-problems |
Core ✅ | Returns a Markdown practice problem prompt. |
calculator-tutor |
Core ✅ | Returns a Markdown tutoring content prompt. |
solve_math_problem |
Stub | Returns a message: "Limited support in SSE demo". |
explain_formula |
Stub | Returns a message: "Limited support in SSE demo". |
calculator_assistant |
Stub | Returns a message: "Limited support in SSE demo". |
calculator://constants |
Core ✅ | Resource for static JSON constants. |
calculator://history/* |
Extended ✅ | Resource for session-specific calculation history. |
calculator://stats |
Extended ✅ | Resource for session-specific usage statistics. |
formulas://library |
Not Implemented | This resource is not included in this example. |
🚀 Getting Started
Prerequisites
- Node.js (v18.x or higher)
- npm or yarn
Installation
# Clone the repository
git clone https://github.com/modelcontextprotocol/mcp-server-examples.git
cd mcp-server-examples/sse
# Install dependencies
npm install
# Build the project
npm run build
Running the Server
# Start the server on port 1923
npm start
# Or, run in development mode with auto-reload
npm run dev
Testing with MCP Inspector
You can interact with the running server using the official MCP Inspector CLI, which understands the two-endpoint SSE transport.
npx @modelcontextprotocol/inspector --cli http://localhost:1923/sse --transport sse
📋 API Usage Examples
The following curl examples demonstrate the required two-terminal interaction.
1. Connect and Establish Session
In your first terminal, establish a persistent connection to the /sse endpoint. This terminal will now display all events sent from the server for your session.
# In Terminal 1: Keep this running to see server-sent events
# The -N flag disables buffering, showing events as they arrive.
curl -N http://localhost:1923/sse
# Server Response:
# event: endpoint
# data: /messages?sessionId=YOUR_UNIQUE_SESSION_ID
Copy the sessionId from the response data. You will need it for the next step.
2. Call a Tool
In a second terminal, use the sessionId you just received to make a POST request to the /messages endpoint.
# In Terminal 2: Send a command
# Replace YOUR_UNIQUE_SESSION_ID with the actual ID from Terminal 1.
curl -X POST 'http://localhost:1923/messages?sessionId=YOUR_UNIQUE_SESSION_ID' \
-H 'Content-Type: application/json' \
-d '{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "calculate",
"arguments": { "op": "multiply", "a": 7, "b": 6 }
}
}'
The server will respond with 202 Accepted in Terminal 2. The actual result of the calculation will appear as a message event in Terminal 1.
🧠 State Management Model
State is ephemeral and scoped to the SSE session. This model is fundamental to the transport's design and limitations.
- Session Registry: As seen in
src/server/index.ts, a globaltransportsobject maps eachsessionIdto its activeSSEServerTransportinstance. This is the core routing mechanism. - In-Memory Isolation: Each new connection to
/ssetriggers thecreateCalculatorServer()factory, which creates a new, isolatedMcpServerinstance. State, such as thecalculationHistoryarray, is local to that instance and not shared between sessions. - No Persistence: When a client disconnects, the
transport.onclosehandler inindex.tsfires, which removes the session from the global registry (delete transports[sessionId]). The server instance and all its associated in-memory state are then garbage collected.
🛡️ Security Model
As a network-exposed service, this transport relies on HTTP-based security patterns.
- Session ID: The
sessionIdgenerated bycrypto.randomUUID()acts as an ephemeral bearer token for the duration of the connection. It authenticatesPOSTrequests to a specific client's event stream. - CORS: The server in
src/server/index.tsis explicitly configured withcors()middleware to allow cross-origin requests, which is essential for browser-based clients. The allowed origin can be restricted for production environments via theCORS_ORIGINenvironment variable. - Input Validation: All incoming tool parameters are rigorously validated against Zod schemas defined in
src/types/calculator.tsto prevent invalid data from causing runtime errors. - No Resumability: While a limitation, the lack of resumability also provides a security benefit: a lost or stolen
sessionIdis only useful as long as the original SSE connection is active, limiting its exposure.
🧪 Testing
This project includes a test suite covering the server's functionality.
# Run all tests (51 passing, 8 skipped expected)
npm test
# Run tests with code coverage report
npm run test:coverage
# Run tests in watch mode for development
npm run test:watch
📚 Official Resources
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.