Ndlovu Code Reviewer
Automates and enhances code review by combining static analysis with AI-powered, human-friendly feedback for uncommitted changes.
Acerca de
This project offers an AI-powered code review assistant designed to address the time-consuming nature and limitations of manual code reviews. By leveraging Model Context Protocol (MCP) tooling, it provides AI assistants with access to a specialized reviewer that automates the busywork of gathering diffs and lint results from local, uncommitted changes. It streams this contextual information to models like Gemini CLI, enabling them to focus on delivering actionable insights. The tool returns structured JSON reviews, seamlessly integrating with MCP-compatible clients to streamline development workflows and improve code quality.
Características Principales
- Connects to MCP clients over stdio using the official SDK
- Runs a hybrid review workflow combining Git diffs and linter findings
- Automatically falls back between ESLint, JSHint, and TypeScript for maximum code coverage
- Safely invokes the Gemini CLI, managing long prompts and timeouts
- Ships with Zod-backed TypeScript types for predictable MCP responses
- 1 GitHub stars
Casos de Uso
- Automating code reviews for local, uncommitted changes to save developer time
- Integrating AI-powered static analysis and contextual feedback directly into IDEs via MCP clients
- Enhancing AI assistant capabilities to provide structured, actionable code quality assessments
