Developed and delivered token usage tracking for embedding operations within the mastra-ai/mastra repository, focusing on the memory module. The work involved updating the saveMessages and recall functions to return an additional usage object, providing visibility into the number of tokens consumed during embedding processes. This enhancement required modifications to core TypeScript interfaces, ensuring optional usage data is surfaced in return types. The implementation was validated through comprehensive unit testing and simulated request flows, supporting full stack development practices. These changes enable more effective monitoring and management of embedding costs, laying the groundwork for future cost governance and optimization efforts.