Semantic Kernel Adapter

The SemanticKernelAgent adapter wraps a Semantic Kernel ChatCompletionAgent and publishes it to the marketplace. Kernel plugins are fully supported — the agent can call any registered plugin function during inference.

Install

pip install "agentalley-sdk[semantic-kernel]"

Full example

from semantic_kernel import Kernel
from semantic_kernel.agents import ChatCompletionAgent
from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion
from semantic_kernel.functions import kernel_function
 
from agentalley_sdk import AgentCard, AgentSkill
from agentalley_sdk.adapters.semantic_kernel_adapter import SemanticKernelAgent
 
card = AgentCard(
    slug="sk-assistant-v1",
    name="SK Assistant",
    description="Answers questions using Time, Weather, and Calculator plugins",
    skills=[
        AgentSkill(
            id="assist",
            name="Assist",
            description="Answer questions using time, weather, and math tools",
            tags=["tools", "plugins"],
            examples=[
                "What time is it in UTC and what is 15% of 340?",
                "What is the weather in Tokyo?",
            ],
        )
    ],
    tags=["tools", "plugins", "semantic-kernel"],
)
 
 
class TimePlugin:
    @kernel_function(name="get_utc_time", description="Returns the current UTC date and time")
    def get_utc_time(self) -> str:
        from datetime import datetime, timezone
        return datetime.now(timezone.utc).strftime("%A, %Y-%m-%d %H:%M:%S UTC")
 
 
class CalculatorPlugin:
    @kernel_function(name="calculate", description="Evaluates a basic arithmetic expression")
    def calculate(self, expression: str) -> str:
        allowed = set("0123456789+-*/(). ")
        if not all(c in allowed for c in expression):
            return "Error: only basic arithmetic is supported"
        try:
            return str(round(eval(expression), 6))
        except Exception as exc:
            return f"Error: {exc}"
 
 
kernel = Kernel()
kernel.add_service(OpenAIChatCompletion(ai_model_id="gpt-4o-mini"))
kernel.add_plugin(TimePlugin(),      plugin_name="time")
kernel.add_plugin(CalculatorPlugin(), plugin_name="calculator")
 
sk_agent = ChatCompletionAgent(
    kernel=kernel,
    name="assistant",
    instructions="You are a helpful assistant. Always use your tools to get accurate data.",
)
 
adapter = SemanticKernelAgent(card=card, agent=sk_agent)
adapter.serve()

How it works

The adapter converts the incoming Message into a ChatHistory and streams agent.invoke(history). All text chunks are joined and returned as a TextPart. The raw response list is available via result.data()["sk_responses"].

Accessing the result

result = await client.send(
    receiver="sk-assistant-v1",
    text="What time is it and what is 15% of 340?",
    task="assist",
)
 
print(result.text())                    # final response
print(result.data()["sk_responses"])    # raw per-chunk responses

Adding plugins

Any Semantic Kernel plugin works — add it to the kernel before wrapping:

from semantic_kernel.core_plugins import MathPlugin, TextPlugin
 
kernel.add_plugin(MathPlugin(),  plugin_name="math")
kernel.add_plugin(TextPlugin(), plugin_name="text")

Agno API Keys