Access Azure AI Foundry threads & messages via Azure Python SDK

Question

Access Azure AI Foundry threads & messages via Azure Python SDK

桂學文 Kevin Kuei 125

Hi,

I understand that Azure AI Foundry stores agent threads and messages. I would like to ask if there is a Python API available that allows me to retrieve the threads and their corresponding messages? That way, I can access the previous messages and continue the conversation from where it left off.

Regards,

Kevin Kuei

Prashanth Veeragoni 4,930 Reputation points Microsoft External Staff Moderator

2025-06-11T11:56:16.65+00:00
Hi 桂學文 Kevin Kuei,

As of now (June 2025), Azure AI Foundry (especially in the context of agents, chat playgrounds, and orchestration tools) does not have public documentation or support for accessing agent threads and messages via the Azure Python SDK (i.e., azure-ai-* libraries).

However, here’s a breakdown of what's possible and what isn't:

Agent threads/messages are stored internally by Azure AI Foundry.

• These are typically used to maintain context across sessions in a chat-based interface (like Copilot Studio or Foundry Agent Orchestrator).

• But Azure does not yet expose a public REST API or SDK to query or manipulate these threads directly.

Current SDK (Python) like azure-ai-ml, azure-ai-openai, etc., do not provide methods to fetch chat history or agent thread logs.

Workarounds:

Until Azure exposes the API for threads/messages, here are your best alternatives:

Manual Session Management

• You can store messages manually in your app using a database or local file.

• For each message you send/receive from the model, append it to a list or dictionary to simulate a thread.

• When making the next API call, pass the message history to maintain context.

Example:

conversation = [ {"role": "user", "content": "Hi"}, {"role": "assistant", "content": "Hello, how can I help?"} ] response = openai.ChatCompletion.create( model="gpt-4", messages=conversation ) conversation.append({"role": "user", "content": "Continue our last chat..."}) conversation.append({"role": "assistant", "content": response['choices'][0]['message']['content']})

Use Azure OpenAI Chat Completions API with AzureOpenAI SDK

• If your agents are backed by OpenAI models (e.g., GPT-4), you can simulate thread-like behavior using:

pip install azure-ai-openai from azure.identity import DefaultAzureCredential from azure.ai.openai import OpenAIClient endpoint = "https://<your-resource-name>.openai.azure.com" client = OpenAIClient(endpoint=endpoint, credential=DefaultAzureCredential()) chat_messages = [ {"role": "user", "content": "Tell me a joke"}, {"role": "assistant", "content": "Why did the chicken cross the road?..."}, ] response = client.chat.completions.create( model="gpt-4", messages=chat_messages, temperature=0.7 )

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.

**

Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.

Thank you!

Accepted answer

0 additional answers

Your answer

Answer 1

Hello @桂學文 Kevin Kuei ,

Yes, you can get the agents, threads and messages using python sdk.

Check this documentation where you can use azure-ai-agents or azure-ai-projects package.

Below is the sample code using azure-ai-projects package with AIProjectsClient.

from azure.ai.projects import AIProjectClient
from azure.identity import DefaultAzureCredential

project_client = AIProjectClient(endpoint="Foundry-based-project-endpoint", credential=DefaultAzureCredential())

agents = list(project_client.agents.list_agents())
threads = list(project_client.agents.threads.list())
messages = list(project_client.agents.messages.list(threads[0].id))

Above code gets agents, threads and messages based on thread id.

user_messages = [msg for msg in messages if msg.get("role") == "user"]
assistant_messages = [msg for msg in messages if msg.get("role") == "assistant"]

print("User Messages:")
for msg in user_messages:
    print(msg.content)

print("\nAssistant Messages:")
for msg in assistant_messages:
    print(msg.content)

Output:

enter image description here

Customize the above code for you use case.

If the solution helped you do accept and give feedback clicking on yes.

Thank you

Share via

Access Azure AI Foundry threads & messages via Azure Python SDK

0 additional answers

Your answer