Sentinel is a hybrid static analysis and AI-assisted code intelligence tool designed to analyze Python codebases and generate structured, professional-grade code review reports. The system combines deterministic program analysis techniques with controlled AI augmentation to provide accurate, reproducible, and context-aware insights into software quality.
The core of Sentinel is a static analysis engine built using Python’s Abstract Syntax Tree (AST) module. The system parses source code to extract structural elements such as functions, classes, imports, control flow constructs, and call relationships. Based on this parsed structure, Sentinel computes quantitative software engineering metrics including cyclomatic complexity, nesting depth, function length, recursion detection, unused imports, global state usage, and an overall maintainability score.
Unlike traditional AI-based review tools that rely solely on language models, Sentinel separates analytical computation from AI interpretation. All complexity metrics and risk indicators are calculated deterministically through rule-based algorithms. The AI layer receives structured analysis results in JSON format and converts them into well-formatted, professional review summaries, refactoring suggestions, and risk classifications. This ensures that AI enhances clarity and usability without compromising technical accuracy.
The system follows a modular architecture consisting of a parsing layer, analysis engine, AI augmentation module, and CLI interface. Sentinel is designed as a command-line developer tool capable of generating terminal reports, JSON outputs, and Markdown-based review documents. The project emphasizes clean architecture, unit testing, extensibility, and potential CI/CD integration.
This project demonstrates practical knowledge in static program analysis, software metric computation, AST traversal, modular system design, and AI-assisted developer tooling. It aims to bridge traditional static analysis with intelligent interpretation to support maintainable and high-quality software development.