Files

Memory Search with the Responses API

This sample demonstrates how to use the Memory Search tool with a ChatClientAgent using the Responses API directly.

What this sample demonstrates

  • Configuring MemorySearchPreviewTool with a memory store and user scope
  • Using memory search for cross-conversation recall
  • Inspecting MemorySearchToolCallResponseItem results
  • User profile persistence across conversations

Prerequisites

  • .NET 10 SDK or later
  • Microsoft Foundry service endpoint and deployment configured
  • Azure CLI installed and authenticated (az login)
  • A memory store created beforehand via Azure Portal or Python SDK

Set the following environment variables:

$env:AZURE_AI_PROJECT_ENDPOINT="https://your-foundry-service.services.ai.azure.com/api/projects/your-foundry-project"
$env:AZURE_AI_MODEL_DEPLOYMENT_NAME="gpt-5.4-mini"
$env:AZURE_AI_MEMORY_STORE_ID="your-memory-store-name"

Run the sample

dotnet run