📦 Agentic SDLC - AI-Powered Software Development Kit

Professional SDK for AI-powered software development. Clean architecture, extensible design, and production-ready.

Agentic SDLC is a Python SDK that provides a comprehensive framework for building AI-powered development tools. It features a clean public API, modular architecture, and support for custom extensions through plugins.

---

🚀 Quick Start

Installation

Core SDK (without CLI):

pip install agentic-sdlc

With CLI tools:

pip install agentic-sdlc[cli]

Development:

pip install -e ".[dev]"

Basic Usage

from agentic_sdlc import Config, Agent, Workflow, get_logger

# Setup logging
setup_logging(level="INFO")
logger = get_logger(__name__)

# Load configuration
config = Config()

# Create and use agents
agent = Agent(name="Developer", role="implementation")
workflow = Workflow(name="feature-dev", agents=[agent])

# Execute workflow
result = workflow.execute()
logger.info(f"Workflow completed: {result}")

CLI Usage

# Initialize a project
agentic init

# Run a workflow
agentic run my-workflow

# View configuration
agentic config get log_level

---

📚 Architecture

Package Structure

agentic_sdlc/
├── core/              # Core functionality (config, exceptions, logging)
├── infrastructure/    # Automation, bridges, execution engines
├── intelligence/      # Learning, monitoring, reasoning, collaboration
├── orchestration/     # Agents, models, workflows, coordination
├── plugins/           # Plugin system and registry
├── cli/               # Command-line interface (optional)
└── _internal/         # Private utilities

Public API

The SDK exposes a clean public API through the top-level agentic_sdlc module:

# Core
from agentic_sdlc import (
    Config, load_config, get_config,
    AgenticSDLCError, ConfigurationError, ValidationError,
    setup_logging, get_logger
)

# Infrastructure
from agentic_sdlc import (
    WorkflowEngine, ExecutionEngine, LifecycleManager
)

# Intelligence
from agentic_sdlc import (
    Learner, Monitor, Reasoner, Collaborator
)

# Orchestration
from agentic_sdlc import (
    Agent, AgentRegistry, create_agent,
    ModelClient, Workflow, WorkflowBuilder
)

# Plugins
from agentic_sdlc import (
    Plugin, PluginRegistry, get_plugin_registry
)

---

🔧 Configuration

Configuration can be loaded from files, environment variables, or set programmatically:

from agentic_sdlc import Config

# Load from file
config = Config("config.yaml")

# Get values with dot notation
log_level = config.get("log_level")
model_name = config.get("models.openai.model_name")

# Set values
config.set("log_level", "DEBUG")

# Validate configuration
config.validate()

---

🔌 Plugin System

Extend the SDK with custom plugins:

from agentic_sdlc import Plugin, get_plugin_registry

class MyPlugin(Plugin):
    @property
    def name(self) -> str:
        return "my-plugin"
    
    @property
    def version(self) -> str:
        return "1.0.0"
    
    def initialize(self, config: dict) -> None:
        print(f"Initializing {self.name}")
    
    def shutdown(self) -> None:
        print(f"Shutting down {self.name}")

# Register plugin
registry = get_plugin_registry()
registry.register(MyPlugin())

---

📖 Documentation

• Getting Started - Installation and basic usage
• Migration Guide - Upgrading from v2.x to v3.x
• Plugin Development - Creating custom plugins
• API Reference - Complete API documentation
• Architecture - Design and architecture decisions

---

💡 Examples

See the examples/ directory for complete working examples:

• Programmatic SDK usage: examples/programmatic/
• CLI usage: examples/cli/
• Plugin development: examples/plugins/

---

🧪 Testing

Run the test suite:

# All tests
pytest

# Unit tests only
pytest tests/unit/

# Property-based tests
pytest tests/property/

# With coverage
pytest --cov=agentic_sdlc --cov-report=html

---

🔄 What's New in v3.0.0

Breaking Changes

• Reorganized package structure (moved to src/ layout)
• Removed vendored dependencies from lib/ directory
• CLI is now optional (install with [cli] extra)
• Old import paths are deprecated (use new public API)

New Features

• Clean public API with explicit exports
• Plugin system for extensibility
• Comprehensive configuration management
• Type hints throughout the codebase
• Property-based testing for correctness validation

Migration

See MIGRATION.md for detailed migration instructions from v2.x.

---

📄 License

MIT License. See LICENSE for details.

---

Developed by Dao Quang Truong | GitHub