chore: import upstream snapshot with attribution

python/examples/mcp.py

@@ -0,0 +1,34 @@
+import asyncio
+
+from langchain_mcp_adapters.client import MultiServerMCPClient
+from langgraph.prebuilt import create_react_agent
+
+from composio import Composio
+
+composio = Composio()
+
+mcp_config = composio.mcp.create(
+    name="langchain-slack-mcp",
+    toolkits=[{"toolkit": "slack", "auth_config_id": "<auth-config-id>"}],
+)
+
+mcp_server = mcp_config.generate(user_id="<user-id>")
+
+client = MultiServerMCPClient(
+    {
+        "composio": {
+            "url": mcp_server["url"],
+            "transport": "streamable_http",
+        }
+    }
+)
+
+
+async def langchain_mcp(message: str):
+    tools = await client.get_tools()
+    agent = create_react_agent("openai:gpt-4.1", tools)
+    response = await agent.ainvoke({"messages": message})
+    return response
+
+
+mcp_response = asyncio.run(langchain_mcp("Show me 20 most used slack channels"))