gurus

To get an expert code review, this Skill analyzes your repository using eight persona-driven expert perspectives to surface architectural, testing, and maintainability defects. It synthesizes their feedback into a prioritized consensus action plan for items agreed upon by six or more experts.

Automating consensus-based code refactoring involves running an autonomous loop to implement action plans generated from multi-perspective expert reviews. This Skill applies accepted changes as one commit per logical item when explicitly authorized by the user.

Multi-perspective code review surfaces hidden design issues by generating independent evaluations from eight expert personas, such as Martin Fowler and Uncle Bob. It aggregates their findings, counts guru agreement, and highlights actionable defects that meet a six-or-more consensus threshold.

Yes, you can scope an automated code review to a specific directory, file, or commit range. While the default behavior reads all relevant source files to avoid diff bias, it accepts explicit scope parameters to focus the architectural and style evaluations.

Yes, multi-perspective code review works across multiple programming languages. The Skill evaluates source files throughout your repository without language restrictions, applying expert testing and style criteria to all relevant code.

You should avoid using autonomous consensus loops for code refactoring when you need strict manual oversight of architectural changes. Instead, use the manual approval mode to review action plans before accepting changes as individual logical commits.