Suraj Sharma

@AlexisMarasigan this is a critical scope issue in `get_session()`. The event iteration loop (lines 183-184) is placed outside the `async with self._get_api_client()` block that ends at line 165, so the...

Confirmed this is a regression in v1.17.0. The issue is in `_session_util.py` line 47 where `extract_state_delta()` explicitly filters out temp: prefixed variables with `elif not key.startswith(State.TEMP_PREFIX)`. When AdkApp passes authorizer...

@rghv-cwx To get user email from the access token, use Google's userinfo endpoint. In your before_tool_callback or tool: ```python import httpx async def get_user_email(access_token: str) -> str: async with httpx.AsyncClient()...

@rghv-cwx Here's a working example for Agent Engine + Gemini Enterprise (requires ADK >= 1.17 and aiplatform >= 1.126.0): ```python from google.adk.agents import LlmAgent def before_tool_callback(callback_context, tool): # auth_id is...

@rghv-cwx Thanks for sharing your blog post. That's a great resource for the community - the step-by-step guide with dynamic filtering using user identity is exactly what others implementing Agentspace...

@rghv-cwx Here are a few channels where your implementation guide would be valuable: 1. **ADK GitHub Discussions** - Post in the [google/adk-python Discussions](https://github.com/google/adk-python/discussions) under "Show and tell" to share with...

@miyannishar you can actually differentiate rubrics per turn Instead of putting all rubrics in `conversation_level_rubrics`, add turn-specific rubrics directly to each invocation using the `rubrics` field. For example, in your...

the current `rubric_based_tool_use_quality_v1` implementation doesn't support per-invocation rubrics. All rubrics from the metric's criterion are evaluated against all turns (see `rubric_based_tool_use_quality_v1.py:193-194`). The inconsistency comes from LLM sampling variance (5 samples...

Thanks for the detailed testing - you're right that this goes beyond configuration. The issue is that `format_auto_rater_prompt` (line 193-194) uses all rubrics from the criterion for every turn without...

@guillaumeblaquiere The issue is in `credential_manager.py` lines 186-187 where `_load_existing_credential()` returns the cached `self._auth_config.exchanged_auth_credential` without checking the current user context from `callback_context`. This causes credentials from the first user to...