🤖 Vibe PR Reviewer

中文文档 | English

An intelligent Pull Request code review bot based on Gemini AI that automatically performs in-depth code reviews on GitHub PRs and provides constructive improvement suggestions.

Features

• 🤖 Intelligent code review based on Google Gemini AI
• 📝 Automatic generation of detailed review reports with issue identification and improvement suggestions
• 🔄 Support for automatic Webhook triggers and manual comment triggers (/review)
• 📁 Smart file context analysis for more accurate review opinions
• 🏷️ Automatic addition of review labels
• ⚡ Support for retry mechanisms and error handling
• 🛠️ Flexible environment variable configuration

Demo

View actual operation results: Example PR Review

Installation and Deployment

1. Clone the Project

git clone <repository-url>
cd pr-review-bot

2. Install Dependencies

pip install -r requirements.txt

3. Environment Variable Configuration

Copy the example configuration file and fill in actual values:

cp env.example .env

Then edit the .env file, or directly set the following environment variables:

Required Environment Variables

# GitHub Personal Access Token (requires repo permissions)
# How to get: GitHub Settings → Developer settings → Personal access tokens → Tokens
# Required permissions: Pull requests, Issues
GITHUB_TOKEN=your_github_token

# Google Gemini API Key
GEMINI_API_KEY=your_gemini_api_key

Optional Environment Variables

# AI model configuration
AI_MODEL_NAME=gemini-2.5-pro                    # Default: gemini-2.5-pro

# Review configuration
REVIEW_LABEL=ReviewedByUllrAI                   # Default: ReviewedByUllrAI
MAX_PROMPT_LENGTH=200000                        # Default: 200000
INCLUDE_FILE_CONTEXT=true                       # Default: true
CONTEXT_MAX_LINES=400                          # Default: 400
CONTEXT_SURROUNDING_LINES=50                   # Default: 50
MAX_FILES_PER_REVIEW=50                        # Default: 50

# Output language configuration
OUTPUT_LANGUAGE=english                        # Default: english (can be any language like "Chinese", "Japanese", etc.)

# Network and retry configuration
MAX_RETRY_ATTEMPTS=3                           # Default: 3
RETRY_DELAY=2.0                               # Default: 2.0
REQUEST_TIMEOUT=60                            # Default: 60

# Server configuration
PORT=5001                                     # Default: 5001

4. Start the Service

python app.py

The service will start on the specified port (default 5001).

5. GitHub Webhook Configuration

Configure Webhook in your GitHub repository:

1. Go to repository settings → Webhooks → Add webhook
2. Payload URL: http://your-server:5001/webhook
3. Content type: application/json
4. Events: Select Pull requests and Issue comments

Note: Currently no Webhook signature verification is required, no need to configure secret.

Usage

Automatic Review

The bot will automatically review the following events:

• When PR is created (pull_request.opened)
• When PR is updated (pull_request.synchronize)
• When PR is reopened (pull_request.reopened)

Manual Review Trigger

Enter /review in PR comments to manually trigger code review.

Configuration Instructions

Core Configuration

• GITHUB_TOKEN: GitHub Personal Access Token, requires repo permissions
• GEMINI_API_KEY: Google Gemini API key

AI Configuration

• AI_MODEL_NAME: Gemini model name to use
• MAX_PROMPT_LENGTH: Maximum prompt length sent to AI
• INCLUDE_FILE_CONTEXT: Whether to include complete file context for analysis

Context Configuration

• CONTEXT_MAX_LINES: Maximum line limit for complete files
• CONTEXT_SURROUNDING_LINES: Number of context lines for code snippets
• MAX_FILES_PER_REVIEW: Maximum number of files per review

Output Language Configuration

• OUTPUT_LANGUAGE: Specifies the language for AI review comments (default: english). Can be set to any language like "Chinese", "Japanese", "French", etc. If set to "english" or not configured, no language instruction will be added to AI prompts.

Network Configuration

• MAX_RETRY_ATTEMPTS: Number of retries when API calls fail
• RETRY_DELAY: Retry interval time (seconds)
• REQUEST_TIMEOUT: HTTP request timeout (seconds)

API Endpoints

Health Check

GET /health

Returns service status and configuration information.

Webhook Handling

POST /webhook

Receives GitHub Webhook events and handles PR reviews.

Logging

The application uses structured logging to record all operations, including:

• GitHub API calls
• AI model calls
• Review process status
• Error and exception information

Log level is INFO, error logs will include complete stack traces.

Deployment Recommendations

Production Environment

• Use Gunicorn or uWSGI as WSGI server
• Configure reverse proxy (like Nginx)
• Set up appropriate log rotation
• Monitor service health status

Docker Deployment

Create Dockerfile:

FROM python:3.9-slim

WORKDIR /app
COPY requirements.txt .
RUN pip install -r requirements.txt

COPY app.py .
CMD ["python", "app.py"]

Environment Variable Examples

Refer to the env.example file, which contains detailed descriptions and example values for all configuration items.

Troubleshooting

Common Issues

1. Permission errors: Ensure GITHUB_TOKEN has sufficient permissions
2. API limits: Check GitHub API rate limits and Gemini API quotas
3. Network timeouts: Adjust REQUEST_TIMEOUT and retry configuration
4. Memory usage: Large PRs may require adjusting MAX_PROMPT_LENGTH and MAX_FILES_PER_REVIEW

Log Debugging

View application logs for detailed error information:

python app.py 2>&1 | tee app.log