ReasoningEngineSpec missing effectiveIdentity field causes ParseError when creating Agent Engines

[bmigette]

Environment

• google-cloud-aiplatform version: 1.136.0
• google-adk version: 1.24.0
• Python version: 3.12
• OS: Ubuntu Linux (GCE VM)

Description

When creating a new Reasoning Engine using vertexai.agent_engines.create(), the API returns a response containing an effectiveIdentity field that is not defined in the local protobuf definitions, causing a ParseError.

The effectiveIdentity field is documented in the REST API reference as an output-only field, but the Python SDK's proto definitions don't include it.

Steps to Reproduce

import vertexai
from vertexai import agent_engines

vertexai.init(project="my-project", location="us-central1")

# This triggers the error when parsing the API response
engine = agent_engines.create(display_name="test_engine")

Error Message

Failed to parse spec field: Message type "google.cloud.aiplatform.v1.ReasoningEngineSpec" has no field named "effectiveIdentity" at "Operation.response.spec".
 Available Fields(except extensions): "['sourceCodeSpec', 'serviceAccount', 'packageSpec', 'deploymentSpec', 'classMethods', 'agentFramework']".

Expected Behavior

The SDK should be able to parse the API response without errors. The effectiveIdentity field should be added to the ReasoningEngineSpec proto definition.

Additional Context

This is similar to:

• google.protobuf.json_format.ParseError when loading prompt saved via Vertex Prompt Management #5176 - SpeechConfig missing languageCode
• loading prompt error : Failed to parse generationConfig field: #4995 - GenerationConfig missing responseModalities

The pattern is the same: the API returns fields that the SDK's protobuf definitions don't yet support.

[ssingh5bmcd]

Any official timeline for this one?

I am using this workaround, have pasted this before doing anything in my deployment.py

    # Force ignore_unknown_fields=True to bypass ParseError on undocumented fields
    import google.protobuf.json_format as json_format
    _original_parse = json_format.Parse
    
    def _patched_parse(text, message, ignore_unknown_fields=False, descriptor_pool=None):
        return _original_parse(text, message, ignore_unknown_fields=True, descriptor_pool=descriptor_pool)
        
    json_format.Parse = _patched_parse