MAF-Persisting-01-Simple

MAF-Persisting-01-Simple

This sample demonstrates how to persist (serialize) an AgentThread to disk and later reload it to resume a conversation with the same agent type.

What it shows

• Creating an AgentThread and running a prompt
• Serializing the thread JSON to agent_thread.json in the application folder
• Rehydrating the thread by deserializing the JSON and calling agent.DeserializeThread(...)
• Continuing the conversation with the resumed thread

Setup

1. Configure your AI provider credentials (example using dotnet user-secrets):

cd samples/MAF/MAF-Persisting-01-Simple
dotnet user-secrets set "AzureOpenAI:Endpoint" "https://<your-endpoint>.openai.azure.com/"
dotnet user-secrets set "AzureOpenAI:Deployment" "gpt-5-mini"
# az login is required (samples use AzureCliCredential)

1. Build and run:

dotnet build
dotnet run

Notes

• The sample relies on shared helper files (ChatClientProvider.cs, StreamConsoleHelper.cs, ResponseClientProvider.cs) that are linked into the project via csproj compile includes.
• The saved thread file agent_thread.json will be created in the app working directory. Treat this file as sensitive — it contains conversation history.

Troubleshooting

• If you see authentication errors, verify your user-secrets or environment variables are set correctly for your chosen provider.
• If deserialization fails, ensure you are using the same agent type to deserialize the thread.

Related docs

• Lesson guide: 06-MAF/README-PersistingConversations.md
• Microsoft Learn: Persisting and Resuming Agent Conversations

License: MIT