coderabbit-triage

Triage code review findings by categorizing issues by severity (critical, important, minor), analyzing dependencies across files, and generating a structured task plan that specifies execution strategy—parallel or sequential—with fixer instructions, time estimates, and constraints for each issue.

Yes. This skill consumes code review findings in JSON format, performs distribution and overlap analysis to detect dependencies between issues, then outputs a dispatch plan with severity categorization and parallelization guidelines so fixers know which tasks can run concurrently.

A task dispatch plan includes total issue count, breakdown by severity, recommended execution strategy, per-issue task IDs, file locations, line numbers, issue descriptions, fixer instructions with constraints, time estimates, and parallelization guidelines for coordinating multi-file fixes.

The skill assesses dependencies and overlap across files and code regions to decide whether issues are independent or sequential. It outputs a strategy field and parallelization guidelines in each task's fixer instructions, enabling parallel execution when safe.

The skill accepts JSON input from CodeRabbit, containing raw code review findings across multiple files. It parses this structured data to perform severity analysis, dependency mapping, and strategy computation, then outputs a formatted task plan.

Yes. The skill is designed for multi-file codebases. It analyzes issues across files, detects dependencies and overlaps between regions, and produces a coordinated task plan that accounts for cross-file constraints and parallelization opportunities.