💎 fix: Handle usage_metadata in Title Transaction for Gemini Models

[kojinseok-del]

Summary

I fixed a bug where title generation token transactions were silently skipped for Gemini models because the collectedUsage mapping in titleConvo did not account for Gemini's usage_metadata response format.

• Add an else if (item.usage_metadata) branch in the titleConvo method's collectedUsage mapping to extract input_tokens and output_tokens from Gemini's usage_metadata object.
• Resolve an issue where Gemini models return token usage via usage_metadata instead of the usage or tokenUsage keys that the existing code handled, causing the token values to remain undefined and the title generation transaction to be silently dropped.

Change Type

• Bug fix (non-breaking change which fixes an issue)

Testing

• Configure a Gemini model (e.g., gemini-2.0-flash) as the title generation model via titleModel or as the active agent model.
• Send a message to trigger a conversation and title generation.
• Verify that the title generation transaction is recorded in the database with correct input_tokens and output_tokens values (previously these would be undefined, causing the transaction to be skipped).
• Confirm that non-Gemini models (OpenAI, Anthropic, etc.) continue to record title transactions correctly through the existing usage and tokenUsage branches.

Checklist

• My code adheres to this project's style guidelines
• I have performed a self-review of my own code
• My changes do not introduce new warnings

[danny-avila]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Another round soon, please!

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".