Worked on the All-Hands-AI/agent-sdk repository to deliver a Security Risk Prediction feature within the MCP Tool Schema, enabling risk-aware safety assessments for automated tooling. Developed a new schema generation path and converted the tool into a Responses API function tool, ensuring compatibility with security risk considerations. Utilized Python for backend and API development, and implemented comprehensive unit and integration tests to validate both functionality and schema behavior. Addressed a schema generation bug to improve reliability and downstream compatibility. The work enhanced safety risk visibility, increased test coverage, and provided clearer risk scoring for applications relying on automated safety assessments.