ce-review

To orchestrate automated code reviews across multiple files, dispatch parallel specialist reviewers for security, performance, and reliability, then merge structured findings into a unified report with confidence scoring.

An adversarial code review is a final red-team pass that challenges initial findings. It improves pull request quality by catching edge cases missed by standard reviews before merging structured findings.

Yes, you can automatically apply mechanical fixes by routing findings using an autofix class. Safe auto fixes apply directly, while gated auto fixes require confirmation before applying to your diff.

Generate a machine-readable code review report for CI pipelines by running the review in headless or report-only mode. This produces a structured JSON envelope containing deduplicated findings and confidence scores.

Yes, this automated review tool works without external dependencies. It operates independently to enforce confidence thresholds and route findings by autofix class across your pull requests.

Confidence scoring limitations arise when evaluating subjective taste-call decisions. The system routes these lower-confidence findings to advisory or manual classes, preventing unsafe automated fixes in your code review.