Reviewer MCP

An MCP (Model Context Protocol) service that provides AI-powered development workflow tools. It supports multiple AI providers (OpenAI and Ollama) and offers standardized tools for specification generation, code review, and project management.

Features

• Specification Generation: Create detailed technical specifications from prompts

• Specification Review: Review specifications for completeness and provide critical feedback

• Code Review: Analyze code changes with focus on security, performance, style, or logic

• Test Runner: Execute tests with LLM-friendly formatted output

• Linter: Run linters with structured output formatting

• Pluggable AI Providers: Support for both OpenAI and Ollama (local models)

Installation

Configuration

Environment Variables

Create a .env file based on .env.example:

Project Configuration

Create a .reviewer.json file in your project root to customize commands:

Using with Claude Desktop

Add the following to your Claude Desktop configuration:

Using with Ollama

1. Install Ollama: https://ollama.ai

2. Pull a model: ollama pull llama2 or ollama pull codellama

3. Set AI_PROVIDER=ollama in your .env file

4. The service will use your local Ollama instance

Available Tools

generate_spec

Generate a technical specification document.

Parameters:

• prompt (required): Description of what specification to generate

• context (optional): Additional context or requirements

• format (optional): Output format - "markdown" or "structured"

review_spec

Review a specification for completeness and provide critical feedback.

Parameters:

• spec (required): The specification document to review

• focusAreas (optional): Array of specific areas to focus the review on

review_code

Review code changes and provide feedback.

Parameters:

• diff (required): Git diff or code changes to review

• context (optional): Context about the changes

• reviewType (optional): Type of review - "security", "performance", "style", "logic", or "all"

run_tests

Run standardized tests for the project.

Parameters:

• testCommand (optional): Test command to run (defaults to configured command)

• pattern (optional): Test file pattern to match

• watch (optional): Run tests in watch mode

run_linter

Run standardized linter for the project.

Parameters:

• lintCommand (optional): Lint command to run (defaults to configured command)

• fix (optional): Attempt to fix issues automatically

• files (optional): Array of specific files to lint

Development

End-to-End Testing

The project includes a comprehensive e2e test that validates the full workflow using a real Ollama instance:

1. Install and start Ollama: https://ollama.ai

2. Pull a model: ollama pull llama2

3. Run the test: npm run test:e2e

The e2e test demonstrates:

• Specification generation

• Specification review

• Code creation

• Code review

• Linting

• Test execution

All using real AI responses from your local Ollama instance.

License

MIT