GitLab AI Code Reviewer

An AI-powered code review tool that automatically analyzes GitLab merge requests and provides intelligent feedback using OpenAI's GPT models.

Features

• Automated code review for GitLab merge requests
• Intelligent feedback using OpenAI's GPT models
• Multiple review strategies (Standard and Security)
• Type-safe Python implementation
• Comprehensive test coverage

GitLab Integration Setup

1. Create a GitLab Bot Account (recommended):

   • Create a new GitLab account for the bot
   • This account will be used to post review comments on merge requests

2. Create a GitLab Access Token:

   • Go to Settings > Access Tokens in your GitLab instance
   • Create a new token with the following permissions:
     • api (API access)
     • read_user (Read user information)
     • read_repository (Read repository)
     • write_discussion (Post comments)
   • Save the token securely, you'll need it for configuration

3. Configure Project-level Variables:

   • Go to Settings > CI/CD > Variables
   • Add the following variables:
     • OPENAI_API_KEY: Your OpenAI API key
     • GITLAB_TOKEN: The access token created earlier
     • Mark them as Protected and Masked for security

Using the Reviewer in Your Project

Step 1: Add Required Variables

In the project you want to be reviewed, go to Settings > CI/CD > Variables and add:

• OPENAI_API_KEY: Your OpenAI API key
• GITLAB_TOKEN: The GitLab access token from setup step 2 Mark both as Protected and Masked for security.

Step 2: Create GitLab CI Configuration

Add this configuration to your project's .gitlab-ci.yml:

ai-review:
  image: python:3.11-slim
  variables:
    GIT_STRATEGY: clone
  script:
    - pip install git+https://gitlab.com/leonj2-pub/ai-reviewer-gitlab.git#egg=ai-reviewer-gitlab
    - python -m ai_reviewer
  rules:
    - if: $CI_PIPELINE_SOURCE == "merge_request_event"

Step 3: Configure Review Settings (Optional)

Create .ai-reviewer.yml in your project root to customize the review:

review_strategies:
  - standard  # Basic code review
  - security  # Security-focused review

# Optional settings
settings:
  max_files_per_review: 10
  excluded_files:
    - "*.md"
    - "*.txt"
  review_comment_prefix: "🤖 AI Review:"

The reviewer will now automatically run on all merge requests and add comments based on the AI analysis.

Local Development

Prerequisites

• Python 3.11 or higher
• GitLab account and API token
• OpenAI API key

Installation

1. Clone the repository:

git clone <repository-url>
cd ai-reviewer-gitlab

2. Create and activate a virtual environment:

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Set up git hooks:

make setup

This will configure git to use the project's hooks, which include:

• A pre-push hook that runs tests before allowing a push to proceed

Configuration

Set the following environment variables:

export GITLAB_URL="your-gitlab-instance-url"
export GITLAB_TOKEN="your-gitlab-api-token"
export OPENAI_API_KEY="your-openai-api-key"

Usage

Run the main script to start reviewing merge requests:

python main.py

Running Tests

Local Testing

Run tests using pytest:

# Run all tests with coverage
pytest --cov=. --cov-report=term-missing -v tests/

# Run specific test file
pytest tests/test_file_name.py -v

Docker Testing Environment

Build and run tests in a Docker container:

# Using make
make test

# Or manually
docker build -t gitlab-reviewer-test -f Dockerfile.test .
docker run --rm gitlab-reviewer-test

Project Structure

• gitlab_reviewer.py: Main GitLab integration logic
• llm_client.py: OpenAI API client implementation
• review_strategies.py: Different code review strategies
• main.py: Entry point of the application
• tests/: Test suite directory
• .githooks/: Git hooks for development workflow
  • pre-push: Runs tests before allowing a push