	In Browser
	StumbleUpon
	del.icio.us
	Google
	Google Buzz
	reddit
	LinkedIn

	Facebook
	Twitter
	Linkedin
	E-Mail

Agentic AI > MCP Protocol > MCP over HTTP with SSE Transport

MCP over HTTP with SSE Transport

Author: Venkata Sudhakar

Stdio transport works well for single-agent setups where the MCP server runs as a subprocess. But when multiple ADK agents across different machines need the same tools, you need a persistent MCP server accessible over the network. MCP supports HTTP with Server-Sent Events (SSE) as a transport, turning the server into a long-running service that any agent can connect to over HTTP.

FastMCP runs as an HTTP server when you set transport to 'sse'. The agent connects using SseServerParameters with the server URL instead of StdioServerParameters. The SSE connection stays open for the duration of the session, allowing the server to stream tool results back to the agent in real time. This is the recommended transport for production deployments where the MCP server is shared infrastructure.

The below example shows a ShopMax India MCP server running as an HTTP service, with an ADK agent connecting to it over the network.

# shopmax_sse_server.py - MCP server with HTTP/SSE transport
from mcp.server.fastmcp import FastMCP

mcp = FastMCP("shopmax-sse", host="0.0.0.0", port=8080)

PRODUCTS = {
    "laptop": {"name": "Lenovo IdeaPad", "price_rs": 45000, "stock": 42},
    "tv": {"name": "LG 43-inch 4K", "price_rs": 35000, "stock": 15},
    "phone": {"name": "Redmi Note 13", "price_rs": 18000, "stock": 120},
}

@mcp.tool()
def search_products(query: str) -> list:
    # Search products by name or category
    query = query.lower()
    results = [
        {"id": pid, **pdata}
        for pid, pdata in PRODUCTS.items()
        if query in pid or query in pdata["name"].lower()
    ]
    return results if results else [{"message": "No products found"}]

@mcp.tool()
def get_product_details(product_id: str) -> dict:
    # Get detailed info for a specific product
    product = PRODUCTS.get(product_id.lower())
    if not product:
        return {"error": "Product not found"}
    return {"product_id": product_id, **product}

if __name__ == "__main__":
    # Run as HTTP server on port 8080
    print("ShopMax MCP server running at http://localhost:8080/sse")
    mcp.run(transport="sse")

Start the server in a terminal, then connect an ADK agent using SSE transport,

# adk_sse_agent.py - connect to remote MCP server over HTTP
import asyncio
from google.adk.agents import LlmAgent
from google.adk.tools.mcp_tool.mcp_toolset import MCPToolset, SseServerParameters
from google.adk.sessions import InMemorySessionService
from google.adk.runners import Runner
from google.genai import types

async def main():
    # Connect to the running MCP HTTP server
    mcp_tools, exit_stack = await MCPToolset.from_server(
        connection_params=SseServerParameters(
            url="http://localhost:8080/sse"
        )
    )

agent = LlmAgent(
        name="product_agent",
        model="gemini-2.0-flash",
        tools=mcp_tools,
        instruction="You are ShopMax India product assistant. Help customers find products."
    )

session_service = InMemorySessionService()
    runner = Runner(agent=agent, app_name="product_agent", session_service=session_service)
    session = session_service.create_session(app_name="product_agent", user_id="cust1")

queries = [
        "Do you have any laptops available?",
        "What is the price of the LG TV?",
    ]
    for q in queries:
        msg = types.Content(role="user", parts=[types.Part(text=q)])
        async for event in runner.run_async(user_id="cust1", session_id=session.id, new_message=msg):
            if event.is_final_response():
                print("Q:", q)
                print("A:", event.content.parts[0].text)
                print()

await exit_stack.aclose()

asyncio.run(main())

It gives the following output,

Q: Do you have any laptops available?
A: Yes! We have the Lenovo IdeaPad laptop in stock.
Price: Rs 45,000 | Stock: 42 units available. Ready to ship!

Q: What is the price of the LG TV?
A: The LG 43-inch 4K TV is priced at Rs 35,000.
We currently have 15 units in stock. Would you like to order one?

SSE transport makes the MCP server network-accessible, so deploy it behind a reverse proxy like nginx or Cloud Run with HTTPS enabled. Add authentication headers to SseServerParameters using the headers parameter to pass Bearer tokens or API keys with each HTTP request. For high-availability deployments at ShopMax India, run multiple MCP server replicas behind a load balancer - the stateless tool design means any replica can handle any request without session affinity.

Send your comments, suggestions or queries regarding this site to [email protected].