Developed a security-focused feature for the MoonshotAI/kimi-cli repository, introducing configurable prompt injection suppression in YOLO mode to address prompt-based vulnerabilities in KimiSoul-powered applications. Leveraging Python for backend and API development, the solution allows deployments to customize prompt handling behavior through configuration management, directly supporting the CLI security hardening roadmap. The work was delivered as a single, well-documented commit with explicit author attribution, ensuring end-to-end traceability. By enabling applications to reduce risk from prompt injection attacks, this feature improved both reliability and governance for downstream users, reflecting a methodical approach to security and maintainability within the codebase.