MCP Integration
Multi-Agent Frameworks

Multi-Agent Frameworks

These frameworks speak MCP over streamable HTTP. The examples connect to the Route6 client on http://localhost:3000/mcp (binary or container), which needs no auth header. Without a client, use https://gw.route6.me/mcp and add "headers": {"Authorization": "Bearer <key>"} — outbound tools only.

Each agent should have its own Route6 agent and key: its address, bandwidth accounting and audit trail follow the key.

CrewAI

# pip install "crewai-tools[mcp]"
from crewai import Agent
from crewai_tools import MCPServerAdapter
 
server_params = {
    "url": "http://localhost:3000/mcp",
    "transport": "streamable-http",
}
 
with MCPServerAdapter(server_params) as route6_tools:
    researcher = Agent(
        role="Web Researcher",
        goal="Research topics using Route6's web tools",
        backstory="Expert at finding information from a real IPv6 identity",
        tools=route6_tools,
        verbose=True,
    )

LangGraph

# pip install langchain-mcp-adapters langgraph
from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent
 
mcp_client = MultiServerMCPClient({
    "route6": {
        "url": "http://localhost:3000/mcp",
        "transport": "streamable_http",
    }
})
tools = await mcp_client.get_tools()          # the 23 Route6 tools
agent = create_react_agent(model, tools)
result = await agent.ainvoke({"messages": [("human", "What is my public IPv6?")]})

For agents that coordinate with each other, see the Team Plan Quick Start and the Team & Mesh tools reference.