Threat modeling the Kubernetes Agent: from MVC to continuous improvement

What this post added

This post details the iterative evolution of GitLab's threat modeling process, specifically using the Kubernetes Agent as a case study. It outlines the progression from an initial security assessment incorporating threat modeling to a more structured, standalone activity. Key developments include the creation of a dedicated repository for threat models, the refinement of threat modeling templates with specific sections for application decomposition and threat analysis, and the use of issue comments for detailed documentation. The post also highlights specific security findings and mitigations identified during the threat modeling of the Kubernetes Agent, such as issues with public project access and path traversal vulnerabilities, and discusses the evolution of architectural diagrams and data flow diagrams to support the process.