Review-flow – automate 80% of code review so humans focus on the 20% (github.com)

Review-flow has introduced an automated AI-powered code review system called Claude Code, designed to streamline the code review process by handling approximately 80% of the review workload. This tool integrates seamlessly with GitLab and GitHub, allowing developers to push their code and automatically triggering Claude to analyze the changes. It provides real-time tracking, a customizable suite of audit agents tailored to specific project needs, and a dashboard that visualizes progress and performance metrics. Developers can expect iterative follow-ups as Claude reassesses code after fixes, fostering a continuous improvement loop. This advancement is significant for the AI/ML community as it dramatically reduces the time engineers spend on code reviews, enabling them to concentrate on the critical 20% that requires human oversight. The integration of features like the Model Context Protocol (MCP) enhances the structured management of review phases and discussion threads, replacing traditional methods that relied on fragile text parsing. Additionally, the dashboard provides comprehensive insights into both individual developer performance and team-level analysis, making it a valuable tool for optimizing coding practices and fostering collaborative improvement across development teams.