agentsclimarketplace

Get pr review

Skill getpipher/skills/skills/git-tools/skills/get-pr-review

Monorepo of @getpipher/* pi-packages and framework-agnostic Agent Skills — git workflows, quality, workspace, Solana, infra playbooks.

Install
npx -y skills add getpipher/skills --skill get-pr-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Retrieve and analyze PR/MR review feedback to improve code quality

SKILL.md

8.5 KB, ~1.8k tokens by cl100k_base, as published. Nobody here has run it

Bismillah! I'll retrieve and analyze the PR/MR review feedback to help you improve your code based on the reviewer's suggestions.

Arguments provided: $ARGUMENTS

Platform Selection

Flags:

  • --github - Fetch PR review from GitHub using gh CLI (default)
  • --gitlab - Fetch MR review from GitLab using glab CLI

Auto-Detection (default):

  • URLs containing github.com → GitHub
  • URLs containing gitlab.com → GitLab
  • Plain numbers → GitHub (default)

Platform Detection Logic

# Parse platform flag or auto-detect from URL
if [[ "$ARGUMENTS" == *"--gitlab"* ]]; then
    PLATFORM="gitlab"
    CLI_TOOL="glab"
    REVIEW_TYPE="merge request"
elif [[ "$ARGUMENTS" == *"--github"* ]]; then
    PLATFORM="github"
    CLI_TOOL="gh"
    REVIEW_TYPE="pull request"
elif [[ "$ARGUMENTS" == *"gitlab.com"* ]]; then
    PLATFORM="gitlab"
    CLI_TOOL="glab"
    REVIEW_TYPE="merge request"
elif [[ "$ARGUMENTS" == *"github.com"* ]]; then
    PLATFORM="github"
    CLI_TOOL="gh"
    REVIEW_TYPE="pull request"
else
    # Default to GitHub for plain numbers
    PLATFORM="github"
    CLI_TOOL="gh"
    REVIEW_TYPE="pull request"
fi

echo "📌 Platform: $PLATFORM ($REVIEW_TYPE review)"

Using with Thinking Mode: mode: think hard

When executing this command with mode: think hard, I will perform an exceptionally deep and comprehensive analysis:

Advanced Review Analysis

  • Deep Context Understanding: Analyze the entire codebase context, not just the PR changes
  • Cross-Reference Analysis: Connect reviewer feedback to broader architectural patterns and project conventions
  • Hidden Issue Detection: Identify potential issues that reviewers might have missed
  • Performance Impact Assessment: Evaluate performance implications of suggested changes
  • Security Review: Conduct security analysis beyond basic review comments
  • Maintainability Evaluation: Assess long-term code maintainability implications

Reviewer Psychology & Communication

  • Intent Interpretation: Deeply analyze the underlying intent behind each reviewer comment
  • Communication Style Analysis: Understand reviewer preferences and communication patterns
  • Priority Inference: Determine true urgency levels beyond surface-level feedback
  • Relationship Dynamics: Consider team dynamics and reviewer expertise areas

Strategic Implementation Planning

  • Multi-Level Solution Analysis: Explore multiple approaches for each piece of feedback
  • Dependency Mapping: Create detailed dependency graphs for implementation order
  • Risk Assessment Matrix: Evaluate risks and benefits of each potential change
  • Future-Proofing: Consider how changes will affect future development and scalability

Advanced Code Quality Enhancement

  • Pattern Recognition: Identify recurring patterns in feedback for systematic improvements
  • Best Practice Synthesis: Combine reviewer suggestions with industry best practices
  • Optimization Opportunities: Find performance and efficiency improvements beyond the review scope
  • Technical Debt Analysis: Evaluate and address technical debt highlighted by review feedback

This intensive analysis ensures not just addressing review feedback, but elevating the entire codebase quality and team collaboration effectiveness.

Let me:

  1. Identify and fetch the PR/MR review

    • Parse PR/MR number or URL from arguments or detect current branch's PR/MR
    • Use appropriate CLI based on platform:
      • GitHub: gh pr view <number> --json reviews,comments
      • GitLab: glab mr view <number> and glab api for discussions
    • Retrieve all review threads, suggestions, and approval status
  2. Analyze the review feedback comprehensively

    • Identify Claude agent reviews vs. human reviews for specialized handling
    • Categorize feedback by type (bugs, improvements, style, architecture, security)
    • Distinguish between critical issues that must be addressed vs. suggestions
    • Extract specific code locations and recommended changes with line numbers
    • Assess overall review sentiment and approval likelihood
    • Parse Claude agent's structured feedback format and recommendations
  3. Create actionable improvement plan

    • Prioritize feedback items by impact and effort required
    • Generate specific code changes needed for each comment
    • Suggest commit organization for addressing feedback
    • Identify any breaking changes or architectural considerations
  4. Provide clear recommendations

    • If review is positive: Highlight what's working well, celebrate good practices, note minor improvements
    • If Claude agent found the code satisfactory: Summarize positive aspects and readiness for merge
    • If review needs work: Create step-by-step action items with specific code examples
    • For Claude agent feedback: Interpret AI suggestions in context of best practices
    • Suggest response strategy for reviewer comments and how to address concerns
    • Estimate time and complexity for implementing each change
    • Recommend whether to request re-review after changes
  5. Generate implementation roadmap

    • Break down changes into logical, atomic commits with clear messages
    • Suggest order of implementation to minimize conflicts and build failures
    • Identify dependencies between different feedback items
    • Provide templates for updating PR description with changes made
    • Include testing strategy for validating fixes
    • Suggest documentation updates if needed
  6. Execute and verify improvements

    • Apply high-priority fixes immediately where possible
    • Run existing tests to ensure no regressions
    • Validate that feedback has been properly addressed
    • Prepare summary of changes made for reviewer response
  7. Commit and push improvements

    • Stage all implemented changes with git add .
    • Create descriptive commit message summarizing the addressed feedback
    • Use format: "address PR feedback: [specific changes made]"
    • Push changes to remote branch to update the PR
    • Ensure all improvements are visible to reviewers before requesting re-review
  8. Conditional re-review request (SMART EXECUTION)

    IF changes were made:

    • MANDATORY: Post re-review request comment using appropriate CLI
    • Use exact format: "@claude I've addressed the review feedback. Please re-review the changes:"
    • Include specific changes made with file:line references
    • List all modified files
    • End with "Ready for re-review ✅"
    • Verify comment was posted successfully

    IF NO changes were made:

    • Report to user only - NO PR/MR comment needed
    • Provide summary: "Analysis complete: No improvements required"
    • Explain why no changes were needed (already implemented, meets standards, etc.)
    • Confirm PR/MR is ready for merge
    • Note: "No re-review request posted (no changes made)"

    Implementation Logic:

    # Track if any changes were made during the process
    CHANGES_MADE=false
    
    # After each improvement attempt, check for actual changes
    if ! git diff --quiet HEAD~1 HEAD; then
        CHANGES_MADE=true
    fi
    
    # Final decision - only ping @claude if changes exist
    if [ "$CHANGES_MADE" = true ]; then
        if [ "$PLATFORM" = "github" ]; then
            gh pr comment $PR_NUMBER --body "@claude I've addressed the review feedback..."
        else
            glab mr note $MR_NUMBER --message "@claude I've addressed the review feedback..."
        fi
    else
        echo "✅ Analysis complete: No improvements required - reporting to user only"
    fi
    
  9. Provide suggested improvements

    • Based on deep analysis, present additional code quality improvements beyond the review feedback
    • Suggest architectural enhancements, performance optimizations, or maintainability improvements
    • Recommend best practices that could elevate the code further
    • Offer insights on patterns, testing strategies, or documentation that could benefit the codebase

Usage Examples

# Get GitHub PR review (default)
/git:get-pr-review 42

# Get GitHub PR review (explicit)
/git:get-pr-review 123 --github

# Get GitLab MR review
/git:get-pr-review 45 --gitlab

# Auto-detect from GitHub URL
/git:get-pr-review https://github.com/owner/repo/pull/15

# Auto-detect from GitLab URL
/git:get-pr-review https://gitlab.com/owner/repo/-/merge_requests/45

Alhamdulillah, let me fetch and analyze the review feedback to help you create an excellent PR/MR!

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.