Memory Bank - Memory Generation with Files

[furoTmark]

Describe the Bug:
Following the https://github.com/GoogleCloudPlatform/generative-ai/blob/main/agents/agent_engine/memory_bank/tutorial_get_started_with_multimodal_agents_with_memory_bank.ipynb tutorial, I tried to generate memories with files, but they are not appearing. I also tried the REST API, it has the same result. There is no error message, also cannot find any error log, only that it was called, but no memory is generated. (Normal text memories work)

Steps to Reproduce:
Please provide a numbered list of steps to reproduce the behavior:

1. Install vertexai
2. Run the python code for memory generation (see below)
3. wait for completion

Expected Behavior:
A memory appears in the Memory section of the Vertex AI Agent Engine.

Observed Behavior:
Finishes without any errors shown, also no error on log explorer

Environment Details:

• ADK Library Version (pip show google-adk): 1.26.0
• Desktop OS:** [e.g., macOS, Linux, Windows]: Windows
• Python Version (python -V): 3.12

Model Information:

• Are you using LiteLLM: No
• Which model is being used: (e.g., gemini-2.5-pro) N/A

---

Minimal Reproduction Code:
Please provide a code snippet or a link to a Gist/repo that isolates the issue.

client.agent_engines.memories.generate(
    name=AGENT_ENGINE_NAME,
    scope={"app_name": APP_NAME},
    direct_contents_source={
        "events": [
            {
                "content": {
                    "role": "user",
                    "parts": [
                        {"text": ("This is ...")},
                        {
                            "file_data": {
                                "file_uri": "gs://.....jpg",
                                "mime_type": "image/jpeg",
                            }
                        },
                    ],
                }
            }
        ]
    },
    config={"wait_for_completion": True},
)

How often has this issue occurred?:

• Always (100%)

[llalitkumarrr]

Hello @furoTmark ,

From your input it looks like you want to update long term memory, we need to reproduce this to analyze this better for this could you please share which vertexai version you are using?

Also could try an alternative approach and see if it works for you :

1. Create an agent and deploy it on agent engine.
2. Now get agent engine ID.
3. Initiate VertexAiMemoryBankService as:

def my_memory_builder():
    print("--- Memory Builder Called! ---")
    return VertexAiMemoryBankService(
        project=PROVIDE_YOUR_PROJECT_ID,
        location=LOCATION,
        agent_engine_id="AGENT_ENGINE_ID_TO_BE_USED",
    )

4. Update Adk.app to include memory_service_builder:

app = agent_engines.AdkApp(
    agent=my_agent,
    memory_service_builder=my_memory_builder,
    enable_tracing=True,
)

5. Update agent in agent engine:

remote_app = agent_engines.update(agent_engine=app,
                                  resource_name="name",
                                  requirements=[
        "google-cloud-aiplatform[adk,agent_engines]"   
    ]
                                  )

The above procedure will update your agent deployed in agent in engine to use VertexAiMemoryBankService.

To save long term memory create tools to save and retrieve memory and provide those tools to agnet:

example:

async def save_fav_color(tool_context: ToolContext, color: str) -> None:
    logger.info("Starting tool save_fav_color")
    try:
        tool_context.state["user_favorite_color" ] = color
        session = tool_context._invocation_context.session
        await tool_context._invocation_context.memory_service.add_session_to_memory(
            session
        )
        return {
            "success": True,
            "message": "saved color preference.",
            "color": color,
        }
    except Exception as e:
        logger.exception(f"Failed to save to memory: {e}")
        return {"success": False, "error": str(e)}

    



async def get_color_from_preference(tool_context: ToolContext) -> str:
    logger.info("Starting tool get_color_from_preference")
    memory_service = tool_context._invocation_context.memory_service
    if memory_service:
        logger.info("Memory service found in the invocation context.")
        logger.info(type(memory_service))
        try:
            logger.info("Attempting to search memory for the favourite coloer.")
            result = await tool_context.search_memory("user_favorite_color" )
            logger.info(f"Memory search result: {result}")
            return {"success": True, "favorite_color": result}
        except Exception as e:
            logger.exception(f"Memory search failed: {e}")
            return {"success": False, "error": str(e)}
    else:
        logger.info("No memory service available in the invocation context.")
        return {"success": False, "error": "No memory service available."}


my_agent = Agent(
    name="MemoryTester",
    model="gemini-2.5-flash",
    tools=[save_fav_color, get_color_from_preference],
    instruction="Test the memory service type."
)

You can modify these tool as per your requirement.

Trying extracting information from media using Gemini and than save that in into memory using the tools. Kindly check if this works for you and share your inputs. If this does not work we will further into your issue.