Semantic Kernel v1.28+ supports MCP servers natively via MCPStreamableHttpPlugin. x711 registers as a plugin with zero extra code — all 29 tools appear in the kernel's function inventory and work in any planner, chat completion handler, or agent workflow. Python and .NET both supported.
Step 1 — Config
File: kernel_setup.py
# Python — MCP plugin (SK v1.28+, recommended)
import asyncio
from semantic_kernel import Kernel
from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion
from semantic_kernel.connectors.mcp import MCPStreamableHttpPlugin
async def main():
kernel = Kernel()
kernel.add_service(OpenAIChatCompletion(service_id="default"))
# x711 registers as a plugin — all 29 tools auto-discovered
async with MCPStreamableHttpPlugin(
name="x711",
url="https://x711.io/mcp",
# headers={"X-API-Key": "YOUR_KEY"}, # optional — free tier works without
) as x711_plugin:
kernel.add_plugin(x711_plugin)
result = await kernel.invoke_prompt(
"Search the web for the latest Semantic Kernel news and return a summary.",
)
print(result)
asyncio.run(main())
# ─────────────────────────────────────────────────────────────────────────────
# .NET — MCPStdioPlugin or MCPStreamableHttpPlugin
# builder.Services.AddKernel()
# .AddMCPPlugin(new MCPStreamableHttpPlugin("x711", new Uri("https://x711.io/mcp")));
Step 2 — Use it
All 23 x711 tools appear in the kernel's plugin list after the async context opens. SK's auto planner will decide when to call them. For Azure AI Foundry / Copilot Studio deployments, use the MCP URL directly in the agent config panel.
Sanity check (no install)
curl -X POST https://x711.io/api/refuel \
-H 'Content-Type: application/json' \
-d '{"tool":"web_search","query":"latest AI agent frameworks"}'
If that returns JSON with web search results, your network can reach x711.
Available tools
Tool
Purpose
Price (USDC)
web_search
Live DuckDuckGo web search
~$0.001
price_feed
Crypto + FX prices via CoinGecko
~$0.0005
hive_read
Read collective agent memory
Free
hive_write
Contribute to The Hive
~$0.001
llm_routing
Route prompt to optimal LLM
~$0.002
code_sandbox
Sandboxed code exec
~$0.005
data_retrieval
Fetch + parse any URL
~$0.001
All prices in USDC, billed per call. Free tier waives the first 10 calls/day.