corvus-review-r3

You can synthesize raw findings into a GitHub-compatible review document by applying deduplication, false-positive filtering, and severity suppression to generate a structured summary with inline comments.

PR review deduplication merges duplicate findings across multiple analysis passes, retains the highest-severity context, and logs the merges to ensure traceability in the final review output.

Yes, you can automatically generate an action decision to APPROVE, REQUEST_CHANGES, or COMMENT_ONLY based on filtered findings, severity thresholds, and nit-budget enforcement rules.

You can filter false positives and suppress low-severity issues by applying severity thresholds and suppression rules consistent with your PR configuration context to keep the review signal crisp.

The best way to enforce a nit-budget is by synthesizing cross-pass findings into a structured document that automatically filters excessive nitpick comments based on your configured review limits.

Duplicate inline comments occur when raw findings overlap across passes; applying a deduplication phase merges these findings, keeps the highest severity, and renders a single clean review document.