MCP Review Server

MCP Review Server

A comprehensive Model Context Protocol (MCP) server that provides specialized code review capabilities using expert critic frameworks and automated fixing strategies to address identified issues.

Example usagage

The typical workflow involves two steps in your vs code agent chat:

1. Critique: Use a critic tool to analyze code and identify issues
2. Fix: Apply a fixer strategy to address the identified problems

Step 1: Critique Code

For best results start a new chat to clear context, then write a request like

review-design this React component
review-c-memory myfunc()
review-sql-security app_schema.sql

Step 2: Apply Fixes

fix-conservative
fix-comment
fix-zealot

Features

• 36+ Specialized Critics: Each critic focuses on specific aspects of code quality
• 3 Fixing Strategies: Automated methods for addressing critic findings
• Explicit Invocation: Users explicitly choose which critic or fixer to apply to their code
• Expert Frameworks: Uses proven methodologies from the ai-review repository
• Two-Step Workflow: Critique first, then apply appropriate fixing strategy

Setup

1. Clone with submodules to get critic frameworks:

   git clone --recursive https://github.com/jonels-msft/mcp-review.git
   
   # OR if already cloned:
   # git submodule update --init --recursive
   

2. Install dependencies:

   npm install
   

3. Run the server:

   npm start
   

Available Tools

The server automatically loads all critic frameworks from the ai-review/critic directory and fixer strategies from the ai-review/fixer directory.

Critic Tools

All critics follow the naming pattern review-<critic-name> and are organized by category:

Design & UI Critics

• review-design - UI and user experience review
• review-color - Color usage and accessibility
• review-data-visualization - Charts and data presentation

Security Critics

• review-sql-security - SQL injection and database security
• review-sql-data-integrity - Data validation and constraints

Performance Critics

• review-algorithm-performance - Algorithm efficiency analysis
• review-algorithm-correctness - Algorithm correctness verification
• review-logging-performance - Logging overhead analysis

Memory & Resource Management

• review-c-memory - C memory management review
• review-c-error-handling - C error handling patterns
• review-c-portability - Cross-platform C code

Error Handling

• review-general-error-handling - Generic error handling best practices
• review-procedural-error - Procedural error management

Data & Logic

• review-procedural-data - Data structure and flow analysis
• review-procedural-functions - Function design and organization
• review-procedural-flow - Control flow analysis
• review-floating-point - Numerical precision and floating-point issues
• review-datetime - Date/time handling correctness
• review-unicode - Unicode and internationalization

Standards & Compliance

• review-sql-standard-compliance - SQL standard adherence
• review-posix - POSIX compliance and portability
• review-unix-interface - Unix philosophy and interfaces

Development Practices

• review-logging-practices - Logging strategy and implementation
• review-terminology - Naming conventions and clarity
• review-writing - Documentation and comments
• review-estimation - Complexity and effort estimation
• review-benchmarking - Performance measurement practices

Specialized Critics

• review-meson - Meson build system review
• review-probability - Statistical and probabilistic code
• review-relational - Database design and relationships
• review-prompting - AI prompt engineering
• review-logicism - Formal logic and reasoning
• review-taylorism - Workflow and efficiency analysis

Fixer Tools

Each fixer strategy is available as fix-<strategy-name>:

• fix-comment - Add TODO comments marking issues for later fixing
• fix-conservative - Fix only clear-cut, low-risk issues that won't require cascading changes
• fix-zealot - Pick one important issue and fix it comprehensively, even if it requires major reorganization

Development

Running the Server

• npm start - Run the server
• npm run dev - Run in development mode

Testing

• npm test - Run all test suites
• npm run test:mcp - Run MCP protocol validation
• npm run test:comprehensive - Run comprehensive feature tests

Utilities

• npm run status - Display server capabilities and status
• npm run check - Check VS Code integration readiness

Project Structure

mcp-review/
├── src/           # Server implementation
├── tests/         # Test suites
├── scripts/       # Utility scripts  
├── ai-review/     # Critic and fixer frameworks (submodule)
└── README.md      # Project documentation

Credit

Critic frameworks provided by the ai-review repository by Joe Nelson.