code-reviewing

Automated code review analyzes Git diffs across supported languages—Go fully supported, Java/Python/Kotlin/Swift/TypeScript in development—using nine parallel agents to detect bugs, security vulnerabilities, quality issues, and performance bottlenecks simultaneously, outputting a Markdown report with confidence scores.

Yes. Code review scans added or modified files for security vulnerabilities and performance bottlenecks alongside bug detection and quality deficits, assigning confidence scores (0-100) and applying redundancy fusion to reduce false positives before displaying results in the terminal.

Trigger code review in Git workflows by using keywords like 'code review,' 'review code,' or 'PR/MR review.' The system automatically detects your project's language, analyzes only changed files, optionally checks project standards via CLAUDE.md, and outputs findings to terminal and Markdown report.

Go is fully supported for production use. Java, Python, Kotlin, Swift, and TypeScript detection is implemented but pending full feature parity. Automatic language detection identifies your repository's primary languages and applies appropriate analysis rules.

Yes. Nine parallel agents analyze changes concurrently, scaling across multi-language repositories. The Skill processes only modified or added files—not the entire codebase—reducing overhead and enabling efficient reviews even in large projects.

Confidence scoring (0-100) thresholds filter low-confidence findings. Redundancy fusion combines duplicate detections across agents and adds +20 credibility when multiple agents flag the same issue, improving signal-to-noise ratio and reducing false positives.