Developed robust cost tracking for LiteLLM chat outputs within the i-am-bee/bee-agent-framework repository, focusing on accurate cost accounting and improved visibility for chat responses. Leveraged Python to implement a CostBreakdown structure that extracts and stores per-token USD costs, migrating cost references from completion_cost to response_cost for consistency. Enhanced backend logic to capture cost data for both streaming and batched responses, updating merge operations to enforce maximum cost limits and ensure reliable reporting. Applied skills in API integration and data modeling to support precise budgeting and pricing decisions, while maintaining strong code traceability to facilitate future cost-aware routing and analytics.