Skip to content

guinacio/cursor-review

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

cursor-review

A Claude Code plugin marketplace that provides cross-model code review using Cursor CLI and GPT-5.3 Codex High.

Get an independent second opinion on your code from a different AI model, checking correctness, bugs, security, tests, performance, error handling, architecture, and plan compliance.

Prerequisites

  • Claude Code v1.0.33+
  • Cursor installed with CLI access (agent command available in your PATH)

Installation

From GitHub Marketplace

/plugin marketplace add guinacio/cursor-review
/plugin install cursor-review@cursor-review-marketplace

Local Development

claude --plugin-dir ./plugins/cursor-review

Usage

Basic diff review (default)

/cursor-review:cursor-review

Reviews git changes (staged + unstaged) against 8 quality categories.

Full codebase review

/cursor-review:cursor-review full

Reviews the entire project structure and key source files.

Review specific files

/cursor-review:cursor-review src/auth/login.ts

Auto-trigger

Claude can also invoke this skill automatically when it detects situations that would benefit from a second opinion (large PRs, security-sensitive changes, complex refactors).

What Gets Reviewed

Category What It Checks
Code Correctness Logic errors, type errors, race conditions, off-by-one
Bugs Null refs, resource leaks, edge cases, state mutation
Security OWASP Top 10 — injection, XSS, auth, SSRF, secrets
Tests Coverage gaps, brittle tests, weak assertions
Performance N+1 queries, memory leaks, algorithmic complexity
Error Handling Swallowed errors, missing validation, poor messages
Architecture SOLID violations, coupling, circular dependencies
Plan Compliance Auto-detects plans and checks implementation alignment

Output Format

The review produces a structured report with:

  • Summary: scope, files reviewed, issue counts by severity
  • Issues by severity: CRITICAL > HIGH > MEDIUM > LOW, each with file:line, description, and suggested fix
  • Plan compliance (when a plan is detected): alignment assessment, gaps, deviations
  • Positive observations: what the code does well
  • Overall assessment: quality score (1-10), highest risk area, recommendation

Plan Auto-Detection

The skill automatically searches for implementation plans in:

  • .claude/plans/*.md
  • PLAN.md or TODO.md in the project root
  • GitHub PR descriptions (via gh pr view)

If a plan is found, the review includes a plan compliance section comparing the implementation against intended requirements.

License

MIT

About

Cross-model code review plugin for Claude Code — get independent second opinions from GPT-5.3 Codex High via Cursor CLI

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors