Intelligent Code Review Agent

A code review system where multiple specialized AI agents work together to analyze your code, find issues, suggest improvements, and generate detailed review reports. Think of it as having a team of expert code reviewers, each focusing on their area of expertise.

What This Does

I built this to showcase how different agentic AI patterns can work together in a practical application. Instead of one agent trying to do everything, I created specialized agents that each focus on a specific aspect of code review. They work together to give you comprehensive feedback on your code.

Key Features

Here's what makes this system useful:

• Specialized Agents: Each agent focuses on one area - syntax, security, performance, style, best practices, and documentation
• Comprehensive Analysis: Covers everything from basic syntax errors to architectural issues
• Smart Issue Detection: Finds bugs, security vulnerabilities, performance problems, and code smells
• Actionable Suggestions: Not just problems, but specific recommendations on how to fix them
• Detailed Reports: Generates well-formatted reports you can share with your team
• Multiple Languages: Works with Python, JavaScript, TypeScript, and more

How It Works

The system uses six specialized agents, each with their own job:

• Syntax Analyzer: Catches syntax errors and basic structural problems before anything else
• Security Agent: Looks for vulnerabilities like SQL injection, hardcoded secrets, and other security risks
• Performance Agent: Identifies bottlenecks, inefficient algorithms, and optimization opportunities
• Style Agent: Checks your code against style guides and naming conventions
• Best Practices Agent: Reviews design patterns, SOLID principles, and architectural issues
• Documentation Agent: Evaluates how well your code is documented

An orchestrator coordinates all these agents, collects their findings, and puts everything together into a comprehensive report.

Tech Stack

• LangChain: Agent framework
• Gemini API: LLM for code analysis
• Streamlit: Web interface
• Python: Core implementation

Status

Ready for use. All core components implemented and tested.

Quick Start

1. Install dependencies:

pip install -r requirements.txt

2. Set up environment:

   Copy the example env and add your Gemini API key (do not commit the real .env):

   • Unix / macOS:

   cp .env.example .env

   • Windows (PowerShell):

   Copy-Item .env.example .env

   Then edit .env and set GEMINI_API_KEY to your key.

   You can get an API key from Gemini's platform.

3. Run the application:

streamlit run frontend/streamlit_app.py

Docker (Recommended)

Run the application using Docker:

docker build -t code-review-agent .
docker run -p 8501:8501 --env-file .env code-review-agent

Usage

Using the system is straightforward:

1. Input your code - Either paste it directly or upload a file
2. Pick the language - Select Python, JavaScript, TypeScript, or another supported language
3. Choose which agents to run - You can run all of them or just the ones you care about
4. Start the review - Click the button and wait a minute or two while the agents do their work
5. Check the results - Each agent's findings are organized in expandable sections
6. Export the report - Download everything as a Markdown, Text, or JSON file to share with your team

Testing

Run tests without API key:

python test_review_dry_run.py

Run full test with API key:

python test_review.py

Security note: If you accidentally committed secrets (for example a real .env file), remove it from future commits with:

git rm --cached .env
git commit -m "remove sensitive .env"

To purge secrets from repository history you will need to rewrite history (for example using git filter-repo or the BFG Repo Cleaner) — take care, this rewrites commits and will require force-push.

Tech Stack

• LangChain: Agent framework and orchestration
• Gemini API: LLM for code analysis
• Streamlit: Web interface
• Python 3.8+: Core language

Use Cases

• Code Quality Assurance: Automated code review before commits
• Security Auditing: Detect security vulnerabilities automatically
• Performance Optimization: Identify performance bottlenecks
• Best Practices: Ensure code follows best practices
• Documentation: Evaluate and improve code documentation

Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

License

This project is licensed under the MIT License - see the LICENSE file for details.