.Net: New Feature: Feature Request: Add OpenTelemetry GenAI instrumentation sample & docs

[dineshkumarkummara]

---

name: Feature request
about: Suggest an idea for this project

---

Summary

otel-genai-bridges provides a .NET library (SkOtel) that adds both ASP.NET Core middleware and a DelegatingHandler to emit OpenTelemetry Generative AI telemetry for Semantic Kernel apps. Surfacing this as a community sample in the SK docs would give teams a ready-made path to end-to-end GenAI observability.

Key features

• AddSemanticKernelTelemetry extension registers options, middleware, and handler
• Captures prompts/completions/tool invocations, tokens, latency, cost, and RAG retrieval latency
• Bundled sample (dotnet/samples/sk-chat) plus Dockerized Collector → Tempo/Prometheus → Grafana stack

Snippet

builder.Services.AddSemanticKernelTelemetry(options =>
{
    options.System = "azure.ai.openai";
    options.Model = "gpt-4o";
    options.CapturePrompts = true;
    options.CaptureCompletions = true;
});

builder.Services.AddOpenTelemetry()
    .ConfigureResource(r => r.AddService("sk-chat"))
    .WithTracing(t => t
        .AddSource("otel-genai-bridges/semantic-kernel")
        .AddAspNetCoreInstrumentation()
        .AddOtlpExporter());

Proposal

Add a docs/sample entry referencing this instrumentation so developers can adopt OTEL GenAI semantics today. Repository: https://github.com/dineshkumarkummara/otel-genai-bridges