Worked on enhancing Anthropic API tool-use integration within the janhq/jan repository, focusing on improving reliability and consistency in AI-assisted workflows. Developed robust parsing logic for interleaved tool-use messages and implemented wave detection to accurately separate tool and non-tool parts. Introduced early provider-context capture to maintain consistent tool-use behavior across provider switches, reducing context drift and edge-case failures. Utilized TypeScript for both front end and full stack development, emphasizing API integration and chat transport stability. These improvements led to smoother multi-provider support and more reliable sequencing of tool-use, ultimately enabling a more seamless user experience in complex AI interactions.