Prompt file imported from wghglory/my-website (.cursor/commands/pr.md). Copyright stays with the author.

/pr Command

Analyze code changes using MCP server and create a pull request with a well-formatted title and description. The AI will automatically:

• Analyze your code changes
• Generate PR title and description
• Create or update the PR on GitHub using MCP server
• Perform AI code review - analyze the PR code changes against all tech standards
• Submit review comments - leave detailed review comments on GitHub only for problematic issues (good changes should not receive comments)
• Provide you with the PR URL

Usage

/pr

Or with optional base branch:

/pr <base-branch>

Parameters

• base-branch (optional): The base branch to compare against (defaults to main)

Workflow

1. Analyze Code Changes

   • Check current git status and branch
   • Get diff between current branch and base branch
   • Identify changed files, additions, deletions
   • Analyze commit history

2. Generate PR Metadata

   • Create a descriptive PR title following conventional commits format
   • Generate comprehensive PR description including:
     • Summary of changes
     • Files changed
     • Breaking changes (if any)
     • Testing instructions
     • Code review notes

3. Create or Update PR

   • Check if PR already exists for the current branch using list_pull_requests
   • If PR exists:
     • Re-analyze ALL changes (including new commits) between base and head branch
     • Generate updated PR title based on all commits/changes
     • Generate updated PR description reflecting all current changes
     • Update existing PR using update_pull_request with:
       • New title (if changed)
       • New description (reflecting all commits)
   • If PR doesn't exist:
     • Create new PR using GitHub MCP create_pull_request
   • Set base branch (default: main) and head branch (current branch)

4. Perform AI Code Review

   • For new PRs or updated PRs: Review ALL changes in the PR (including new commits)
   • Analyze each changed file for:
     • Code quality and best practices
     • Adherence to project standards and rules
     • Potential bugs or issues
     • Performance considerations
     • Security concerns
     • Consistency with codebase patterns
   • Check for:
     • Linter errors
     • Type safety issues
     • Missing error handling
     • Console.logs or debug code
     • Memory leaks
     • Adherence to project-specific rules
   • Note: When updating an existing PR, review comments will focus on new changes, but the review summary will cover all changes

5. Submit Review Comments

   • IMPORTANT: Only comment on problematic issues, not on good changes
   • Create a pending review using pull_request_review_write with method create
   • Add line-specific review comments using add_comment_to_pending_review only for:
     • Bugs or errors
     • Security vulnerabilities
     • Performance issues
     • Code quality problems
     • Violations of project standards/rules
     • Missing error handling
     • Type safety issues
     • Suggestions for critical improvements
   • Do NOT comment on:
     • Good practices (no need to praise)
     • Well-written code
     • Proper implementations
     • Correct patterns
   • Submit the review using pull_request_review_write with method submit_pending
   • Use COMMENT event (can't approve own PR)
   • If no issues found, submit review with summary only (no line comments)

6. Provide Results

   • Return PR URL
   • Summarize review findings
   • List any issues found

Example

/pr

This will:

• Analyze changes in current branch vs main
• If PR exists: Update title and description to reflect all commits (including new ones)
• If PR doesn't exist: Create PR with title like "feat: Add Cursor mastery blog post and update post rules"
• Generate comprehensive description covering all changes
• Perform code review on all changed files (including new commits)
• Add review comments on GitHub only for problematic issues (good code receives no comments)
• Return PR URL: https://github.com/wghglory/my-website/pull/82

Example with updates:

/pr  # First run - creates PR #82
# ... make more commits and push ...
/pr  # Second run - updates PR #82 with new title/description reflecting all commits

Code Review Standards

The AI will review against:

• Project Rules: Check .cursor/rules/ for project-specific standards
• Code Quality: Best practices, patterns, consistency
• Type Safety: TypeScript/type checking issues
• Performance: Potential performance issues
• Security: Security vulnerabilities
• Testing: Missing tests or test coverage
• Documentation: Missing or unclear documentation
• Linting: Linter errors and warnings

Error Handling

• If no changes detected, inform the user
• If branch doesn't exist on remote, provide instructions to push
• If GitHub MCP server is not connected, show clear error message
• If PR creation fails, show error details
• If review submission fails, show partial results if available

Notes

• The command works best when changes are already committed and pushed
• For uncommitted changes, the command will analyze staged/unstaged changes
• Re-running /pr on the same branch will update the existing PR with:
  • Updated title (if commits suggest a different scope)
  • Updated description (reflecting all commits, not just new ones)
  • Fresh code review of all changes
• Review comments focus only on issues - good code receives no comments to keep reviews concise and actionable
• Review comments are submitted as COMMENT event (not approval) since you can't approve your own PR
• The PR description follows a standard template with sections for Summary, Changes, Breaking Changes, Testing Instructions, and Code Review Notes
• When updating an existing PR, the description will be regenerated to include all commits/changes, ensuring it stays current