coderabbit

Execute CodeRabbit CLI review with critical evaluation, systematic fixes, and project consistency validation.

Arguments

Parse arguments: $ARGUMENTS

• --type: uncommitted (default) | committed | all
• --base: Base branch (intelligently determined if not specified)

Workflow

Phase 1: Base Branch Detection

1. Check current branch: git rev-parse --abbrev-ref HEAD
2. Check unpushed commits: git rev-list --count @{u}..HEAD 2>/dev/null
3. If --base explicit: use that. If ahead of origin: use origin/<branch>. Otherwise: "main"

Phase 2: Execute Review

Run synchronously (takes 7-30+ minutes). CodeRabbit CLI ≥ 0.7 dropped --prompt-only/--type in favor of separate scope flags (plain text is the default output now); map [type] to:

• uncommitted (default) → --uncommitted
• committed → --committed
• all → no scope flag (reviews the full diff against --base)

coderabbit review --base [base]        # type: all
coderabbit review --uncommitted --base [base]
coderabbit review --committed --base [base]

Set timeout to 3600000ms (60 minutes). Inform user about expected duration.

Phase 3: Parse Results

Extract issues from CodeRabbit output:

• File path, line numbers
• Issue type (potential_issue, suggestion)
• Fix description

Create todo list with one item per issue.

Phase 4: Critical Evaluation & Fix

For each issue, critically evaluate before implementing:

1. Is the issue valid? (not a false positive?)
2. Is the fix appropriate? (fits project conventions?)
3. Is the fix necessary? (provides genuine value?)
4. Could the fix cause harm? (breaks functionality?)

Decision framework:

• Apply: Valid and appropriate — implement
• Adapt: Core idea valid, adjust for project — implement modified version
• Skip: False positive or unnecessary — mark completed with justification

Guidelines:

• Skip purely stylistic changes with no functional benefit
• Adapt rather than blindly copy when project conventions differ
• When in doubt, skip and flag for manual review

Phase 5: Validation

After all fixes, validate project consistency:

• Architectural patterns maintained
• Naming conventions consistent
• Dependencies and imports correct
• Cross-reference similar patterns in codebase using Grep

Phase 6: Summary

Report: issues found, fixes applied, items skipped (with reasons), validation results, testing recommendations.