Skip to content

Conversation

@markpollack
Copy link
Member

Summary

  • Extract safety ratings from Gemini response candidates and add to AssistantMessage metadata
  • Add VertexAiGeminiSafetyRating record with HarmCategory, HarmProbability, and HarmSeverity enums
  • Update Google Cloud BOM to 26.72.0
  • Add integration test for safety ratings
  • Add documentation for safety settings and safety ratings

Closes #687

Extract safety ratings from Gemini response candidates and include
them in AssistantMessage metadata. Update Google Cloud BOM to 26.72.0.
@markpollack
Copy link
Member Author

markpollack commented Dec 1, 2025

should be in both 1.1.x. I am not sure about main, since vertex is going away in june 2026 - https://docs.cloud.google.com/vertex-ai/generative-ai/docs/deprecations maybe we just put it in 1.1.x. I think we are missing this functionality in the genai sdk module however. I realize i made this PR against main, thought of this afterwards

@markpollack markpollack added this to the 1.1.1 milestone Dec 1, 2025
@sobychacko
Copy link
Contributor

Merged to main via 2710cab and cherry-picked to 1.1.x.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

VertexAiGeminiChatClient loses information about Safety Ratings when returning ChatResponse

3 participants