🤖 CodeReview Agent

Microsoft Agents League 2026 – Creative Apps Track (GitHub Copilot Chat)

---

📌 What is CodeReview Agent?

CodeReview Agent is an AI-powered chatbot assistant that reviews your code in real time using GitHub Copilot Chat. It detects bugs, security vulnerabilities, and code smells — and suggests clean, refactored alternatives along with ready-to-use unit test stubs.

Designed for developers who want instant, actionable code feedback without waiting for a PR review.

---

✨ Features

Feature	Description
🐛 Bug Detection	Identifies logic errors, null pointer risks, edge cases
🔒 Security Scanning	Flags SQL injection, hardcoded secrets, unsafe I/O
💡 Refactoring Tips	Suggests cleaner, more idiomatic rewrites
🧪 Test Stubs	Generates pytest/unittest stubs automatically
📂 File Review	Review any local file directly from the CLI
💬 Chat Mode	Conversational REPL — ask follow-up questions
🔁 CI Integration	Single-shot mode for use in GitHub Actions pipelines

---

🏗️ Architecture

copilot-code-reviewer/
└── agent.py              # Core agent: CodeReviewAgent class + CLI
└── test_agent.py         # pytest test suite (10+ tests)
├── instructions.md # GitHub Copilot repo instructions
└── ci.yml      # CI pipeline (Python 3.10/3.11/3.12)
├── .env.example              # Credential template
├── requirements.txt
└── README.md

How it uses GitHub Copilot

The agent connects to GitHub's Models Inference endpoint (https://models.inference.ai.azure.com) using your GITHUB_TOKEN. This is the production GitHub Copilot Chat API — the same one powering Copilot in VS Code and the CLI.

User Input
    │
    ▼
CodeReviewAgent (Python)
    │  Builds structured prompt with SYSTEM_PROMPT + conversation history
    ▼
GitHub Copilot Chat API  ──►  gpt-4o (or gpt-4o-mini)
    │
    ▼
Structured Review Output
  📋 Summary  |  🐛 Issues  |  💡 Suggestions  |  ✅ Refactor  |  🧪 Tests

---

🚀 Quick Start

1. Clone the repo

git clone https://github.com/pawan1979/codereview-agent.git
cd codereview-agent

2. Install dependencies

pip install -r requirements.txt

3. Set your GitHub token

cp .env.example .env
# Edit .env and add your GITHUB_TOKEN

Getting a GitHub Token: Go to GitHub → Settings → Developer settings → Personal access tokens → Fine-grained tokens → Generate new token. No special scopes needed for Copilot Models access.

4. Run the agent

# Interactive chat mode
python agent.py

# Review a specific file
python agent.py --file path/to/your/code.py

# Review an inline snippet
python agent.py --code "def divide(a,b): return a/b"

# Save review to JSON (great for CI pipelines)
python agent.py --file app.py --output review.json

---

💬 Interactive Mode Demo

╔══════════════════════════════════════════════════════════╗
║        🤖  CodeReview Agent  |  Powered by GitHub Copilot       ║
║          Microsoft Agents League 2026 Submission           ║
╚══════════════════════════════════════════════════════════╝

You > review def get_user(id): return db.execute(f"SELECT * FROM users WHERE id={id}")

Agent >
📋 Summary: Critical Issues

🐛 Issues Found:
1. [CRITICAL] SQL Injection – f-string interpolation in SQL query allows arbitrary SQL execution
2. [MEDIUM]   No error handling – db.execute may throw if connection is lost

💡 Suggestions:
- Use parameterised queries: db.execute("SELECT * FROM users WHERE id=?", (id,))
- Wrap in try/except and return None on failure

✅ Refactored Snippet:
def get_user(user_id: int):
    try:
        return db.execute("SELECT * FROM users WHERE id = ?", (user_id,)).fetchone()
    except Exception as e:
        logging.error(f"DB error: {e}")
        return None

🧪 Test Stubs:
def test_get_user_returns_none_on_invalid_id():
    assert get_user(-1) is None

You > explain

Agent > This function fetches a user record from the database by ID. The original
version is dangerous because it builds the SQL query by directly embedding
user input into the string — a classic SQL injection vulnerability...

---

🧪 Running Tests

pip install pytest pytest-cov
pytest  -v --cov=src

---

🔧 CI / CD Integration

Use CodeReview Agent in your GitHub Actions pipeline:

- name: AI Code Review
  env:
    GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
  run: |
    python agent.py --file src/my_module.py --output review.json
    cat review.json

---

🌐 Environment Variables

Variable	Required	Description
GITHUB_TOKEN	✅ Recommended	GitHub PAT for Copilot Models API
COPILOT_MODEL	Optional	Model to use (default: gpt-4o)

---

📋 Submission Details

Field	Value
Challenge	Microsoft Agents League 2026
Track	Creative Apps – GitHub Copilot (Chat)
Dates	Feb 16 – Feb 27, 2026
Tech Stack	Python 3.10+, OpenAI SDK, GitHub Copilot Models API
License	MIT

---

📄 License

MIT © 2026 – Built for Microsoft Agents League