Semantic Kernel (tracing)

Microsoft Semantic Kernel is a Python SDK for composing agents, plugins, prompts, and model services. Respan enables Semantic Kernel diagnostics and normalizes kernel function, chat completion, and tool spans into Respan traces.

This tracing setup uses your model provider credentials directly. To route model calls through Respan instead, use the Semantic Kernel gateway setup.

Set up Respan

Create an account at platform.respan.ai and grab an API key.

Run npx @respan/cli setup to set up with your coding agent.

Use Respan Gateway

See Semantic Kernel gateway setup to route model calls through the Respan gateway.

Example projects

Python examples

Setup

Install packages

$ pip install respan-ai respan-instrumentation-semantic-kernel semantic-kernel openai

Set environment variables

$ export OPENAI_API_KEY="YOUR_OPENAI_API_KEY"
$ export RESPAN_API_KEY="YOUR_RESPAN_API_KEY"

OPENAI_API_KEY is used by Semantic Kernel for direct model requests. RESPAN_API_KEY exports traces to Respan.

Initialize and run

1 import asyncio
2 import os
3 
4 from respan import Respan, workflow
5 from respan_instrumentation_semantic_kernel import SemanticKernelInstrumentor
6 from semantic_kernel import Kernel
7 from semantic_kernel.connectors.ai.open_ai import OpenAIChatCompletion
8 from semantic_kernel.connectors.ai.open_ai import OpenAIChatPromptExecutionSettings
9 from semantic_kernel.functions import KernelArguments
10 
11 respan = Respan(
12     app_name="semantic-kernel-demo",
13     instrumentations=[SemanticKernelInstrumentor()],
14 )
15 
16 kernel = Kernel()
17 kernel.add_service(
18     OpenAIChatCompletion(
19         service_id="chat",
20         ai_model_id="gpt-4o-mini",
21         api_key=os.environ["OPENAI_API_KEY"],
22     )
23 )
24 
25 @workflow(name="semantic_kernel_chat")
26 async def run_chat() -> str:
27     result = await kernel.invoke_prompt(
28         "Reply in one sentence: what does Semantic Kernel help developers build?",
29         arguments=KernelArguments(
30             settings=OpenAIChatPromptExecutionSettings(
31                 service_id="chat",
32                 temperature=0.2,
33                 max_tokens=80,
34             )
35         ),
36     )
37     return str(result)
38 
39 async def main() -> None:
40     print(await run_chat())
41 
42 asyncio.run(main())
43 respan.flush()
44 respan.shutdown()

View your trace

Open the Traces page to see kernel function, chat completion, plugin/tool, and error spans.

Configuration

Parameter	Type	Default	Description
`api_key`	`str \| None`	`RESPAN_API_KEY`	Respan API key used to export traces.
`base_url`	`str \| None`	`RESPAN_BASE_URL`	Optional Respan trace export API URL.
`instrumentations`	`list`	`[]`	Include `SemanticKernelInstrumentor()` to activate Semantic Kernel tracing.
`customer_identifier`	`str \| None`	`None`	Default customer identifier for all spans.
`metadata`	`dict \| None`	`None`	Default metadata attached to all spans.
`environment`	`str \| None`	`None`	Environment tag, for example `"production"`.

What gets traced

Semantic Kernel operation	Respan span	Notes
Kernel function invocation	`task`	Captures function name, arguments, result, and failures.
Chat or text completion	`chat` or `text`	Captures prompt/completion content, model, and token usage when Semantic Kernel exposes it.
Plugin tool execution	`tool`	Captures plugin function name, arguments, result, and errors.

Attributes

Set defaults at initialization. These apply to every Semantic Kernel span emitted in the active process.

1 from respan import Respan
2 from respan_instrumentation_semantic_kernel import SemanticKernelInstrumentor
3 
4 respan = Respan(
5     instrumentations=[SemanticKernelInstrumentor()],
6     customer_identifier="user_123",
7     metadata={"service": "semantic-kernel-api", "version": "1.0.0"},
8     environment="production",
9 )

Pass capture_content=False to SemanticKernelInstrumentor() when you want diagnostics without prompt and completion payload content.