DeployMate-II: A Deterministic and AI-Assisted Release Risk Gate for CI/CD Pipelines
Year: 2026
Type: preprint
Abstract: DeployMate-II: A Deterministic and AI-Assisted Release Risk Gate for CI/CD Pipelines Abstract DeployMate-II is a prototype DevSecOps framework designed to improve the reliability, security, and explainability of deployment decisions in Continuous Integration and Continuous Delivery (CI/CD) pipelines. The framework uses a deterministic risk-scoring engine to evaluate build status, automated test results, vulnerability findings, recent deployment failures, and optional infrastructure telemetry. Based on predefined scoring rules and mandatory blocking conditions, the system produces three possible decisions: APPROVE, REVIEW, or BLOCK. The architecture incorporates a FastAPI backend, PostgreSQL database, React dashboard, JWT-based authentication, and a Jenkins-oriented command-line gatekeeper. An optional Amazon Bedrock advisory component provides evidence-based explanations and remediation recommendations without overriding the deterministic risk engine. Functional evaluation across five representative scenarios demonstrates the expected behavior of the scoring policy, including approval of clean releases, review of elevated-risk deployments, and blocking of critical vulnerabilities or failed builds. The project demonstrates an approach to transparent and auditable deployment governance. Further evaluation is required to establish its effectiveness in production environments. Keywords: DevSecOps, CI/CD, Deployment Risk Assessment, Jenkins, Trivy, FastAPI, PostgreSQL, Automated Deployment, AI-Assisted Decision Support.
Institution: MIT Art, Design and Technology University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7236598315
