Portfolio: Code Review Crew

Multi-agent AI code review system built with CrewAI. Four specialised agents collaboratively analyse codebases and produce consolidated review reports with actionable findings.

UI:

Agents

Agent	Focus
🛡️ Security Auditor	Hardcoded secrets, injection risks, vulnerable dependencies (via pip-audit + OSV)
📐 Code Quality Analyst	Complexity metrics (AST analysis), DRY violations, naming, type hints
⚡ Performance Reviewer	Inefficient algorithms, blocking I/O, memory patterns, N+1 queries
📝 Documentation Reviewer	Docstring coverage, README accuracy, inline comments, API docs

Features

• Multi-agent orchestration — sequential CrewAI pipeline with shared memory
• Structured output — Pydantic models with severity-rated findings
• Dual input — accepts GitHub URLs (auto-clones) or local file paths
• Custom tools — AST analyser for objective metrics, pip-audit for real CVE data
• Dual LLM support — OpenAI or local Ollama inference
• Streamlit UI — score cards, expandable findings, JSON export
• CLI mode — headless execution with JSON output

Example Output

See examples/sample_review_report.json for a sample review report.

Quick Start

git clone https://github.com/benwalkerai/Portfolio_CodeReviewCrew.git
cd Portfolio_CodeReviewCrew
cp .env.example .env
# Edit .env with your API key
uv sync

Streamlit UI

uv run streamlit run src/ui/interface.py

CLI

uv run python main.py /path/to/repo -o report.json
uv run python main.py https://github.com/user/repo -o report.json

Docker

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

Configuration

Variable	Default	Description
LLM_PROVIDER	openai	openai or ollama
OPENAI_API_KEY	—	Required if using OpenAI
MODEL_NAME	gpt-4o-mini	Model to use
OLLAMA_BASE_URL	http://localhost:11434	Ollama endpoint
VERBOSE	true	Agent verbose logging

Project Structure

Portfolio_CodeReviewCrew/
├── src/
│   ├── agents/           # Agent definitions (4 specialists)
│   ├── tasks/            # Task definitions per agent
│   ├── tools/            # Custom CrewAI tools
│   │   ├── file_reader.py
│   │   ├── ast_analyser.py
│   │   ├── dependency_checker.py
│   │   └── repo_loader.py
│   ├── ui/               # Streamlit interface
│   ├── config.py         # LLM provider configuration
│   ├── models.py         # Pydantic output schemas
│   └── crew.py           # Crew orchestration
│   ├── tests/            # Unit tests
├── main.py               # CLI entry point
├── Dockerfile
└── pyproject.toml

Tech Stack

Python, CrewAI, Streamlit, Pydantic, pip-audit, AST module

License

MIT

Version: 1.0.0
Author: Ben Walker