PyJunk - AI Engineering System

An AI Engineering system project using Azure AI, AutoGen, and Model Context Protocol (MCP) for intelligent agent-based automation.

Project Structure

├── src/                    # Source code
│   ├── __init__.py        # Package initialization
│   └── main.py            # Main application entry point
├── tests/                  # Test files
│   ├── conftest.py        # Test configuration
│   └── test_example.py    # Example tests
├── scripts/               # Utility scripts
│   └── restore.sh         # Restore script
├── htmlcov/              # Coverage reports (generated)
├── .devcontainer/        # Development container configuration
├── pyproject.toml        # Project configuration and tool settings
├── poetry.lock           # Dependency lock file
├── Makefile              # Build automation
└── README.md             # This file

Key Dependencies

Core AI/ML Libraries

• Azure AI Projects (^1.0.0) - Azure AI project management
• Azure AI Agents (>=1.2.0b2) - Azure AI agent framework
• AutoGen AgentChat (^0.7.4) - Multi-agent conversation framework
• AutoGen Ext with MCP (^0.7.4) - Extended AutoGen with Model Context Protocol
• OpenAI (^1.109.0) - OpenAI API client
• MCP Server Fetch (^2025.4.7) - Model Context Protocol server for web fetching

Infrastructure & Monitoring

• Azure Identity (^1.25.0) - Azure authentication
• OpenTelemetry SDK (^1.37.0) - Observability and tracing
• Azure Core Tracing OpenTelemetry - Azure telemetry integration

Utilities

• Pydantic Settings (^2.10.1) - Configuration management
• TikToken (^0.11.0) - Token counting for AI models

Development Environment

This project uses a comprehensive VS Code Dev Container with:

Pre-installed Tools

• Python 3.12 with Poetry for dependency management
• Node.js 20 for additional tooling
• Azure CLI with Azure Functions Core Tools
• GitHub CLI for GitHub integration
• Docker-in-Docker for containerized development
• Terraform 1.13.2 for infrastructure as code
• Git built from source

Development Tools

• Ruff - Fast Python linter and formatter
• pytest - Testing framework with coverage
• mypy - Static type checking
• VS Code Extensions - Comprehensive extension pack for Python, Azure, and more

Tool Configuration

All tool configurations are centralized in pyproject.toml:

Ruff (Linting & Formatting)

• Line length: 240 characters
• Target: Python 3.12
• Comprehensive rule set including security (bandit), complexity (mccabe), and import sorting
• Per-file ignores for test files

Testing (pytest)

• Auto-discovery of test files
• Coverage reporting with HTML output
• Async test support
• Coverage threshold: 80%
• Source tracking from src/ directory

Type Checking (mypy)

• Python 3.12 compatibility
• Reasonable strictness with helpful error reporting
• Missing import tolerance for rapid development

Coverage

• Source tracking with exclusions for test and virtual environment files
• HTML and terminal reporting
• Excludes common non-testable code patterns

Getting Started

Prerequisites

• VS Code with Dev Container extension
• Docker Desktop

Development Setup

1. Open in Dev Container: VS Code will automatically detect and offer to reopen in container

2. Environment Setup: Container will automatically run post-create scripts

3. Install Dependencies:

   poetry install

4. Activate Virtual Environment:

   poetry shell

Development Workflow

# Run tests with coverage
pytest

# Format code
ruff format .

# Lint code
ruff check .

# Type checking
mypy src/

# Run all checks (if using Makefile)
make test
make lint
make format

Environment Variables

The dev container sets:

• PYTHONPATH to workspace folder for proper imports
• WORKSPACE to workspace folder for scripts
• Python interpreter path to .venv/bin/python

VS Code Configuration

The dev container includes optimized VS Code settings for:

• Python development with intelligent code analysis
• Automatic formatting and import organization on save
• Integrated terminal with Poetry environment activation
• Azure and GitHub integration
• Comprehensive extension pack pre-installed