Minimal samples showing how to instrument third‑party agent frameworks for observability with Azure Application Insights (via OpenTelemetry).
- Focus: high‑signal, compact examples of tracing agent runs and tool calls.
- Providers covered: AWS Bedrock, Google Vertex AI.
- Each folder has its own short README with setup and run steps.
aws/— Bedrock + LangGraph + AgentCore sample with Azure tracing. Seeaws/README.md.gcp/— Vertex AI LangChain agent samples with Azure tracing. Seegcp/README.md.azure/— Microsoft Agent Framework sample for Azure Container Apps with Azure Monitor tracing. Seeazure/README.md.
- Pick a provider folder and follow its README.
- Requirements are listed per folder (e.g.,
requirements.txt). - To deploy the Azure sample end-to-end with Azure Developer CLI, run
azd upfrom the repo root after configuringenv/.env.sample.
POST https://us-west1-aiplatform.googleapis.com/v1/projects/ninhu-project1/locations/us-west1/reasoningEngines/1304319857705091072:query
Body:
{
"input": { "input": "What is the exchange rate from US dollars to SEK today?" }
}Headers:
Authorization: Bearer <token>Get token:
gcloud auth application-default print-access-tokenPOST https://bedrock-agentcore.us-west-2.amazonaws.com/runtimes/arn%3Aaws%3Abedrock-agentcore%3Aus-west-2%3A025211824558%3Aruntime%2Fagentcore_langgraph_agent-1EC4Au3NoU/invocations?qualifier=DEFAULT
Body:
{
"prompt": "What is the exchange rate from USD to EUR today?"
}Auth: AWS Signature Version 4 (SigV4)