📦 Agentic SDLC - Multi-Domain Swarm Agent SDK

Next-generation AI Software Development Lifecycle framework with Swarm Intelligence, RAG-powered Research, and Multi-Provider LLM orchestration.

Agentic SDLC is a modular Python SDK for building autonomous development agents. It moves beyond simple prompt generation into a full E2E pipeline: Domain Detection → Research → Prompt Optimization → Swarm Execution → Self-Learning.

---

🚀 Key Features

• 🧠 Swarm Intelligence: Multi-agent coordination (Developer, Reviewer, Tester, Researcher) with async message bus.
• 🔍 RAG Research: Integrated Knowledge Base using ChromaDB and web-research capabilities.
• 🎯 Domain Engine: Automatic task classification into 7+ technical domains (Frontend, Backend, DevOps, etc.).
• 🧪 Prompt Lab: A/B testing for prompts with structured strategies and AI-powered evaluation.
• 🤖 Multi-Provider LLM: Unified support for Gemini (primary), OpenAI, Anthropic, and Ollama (local).
• 📈 Self-Learning: Continuous improvement engine that analyzes execution patterns and proposes enhancements.

---

📦 quick start

installation

pip install agentic-sdlc[all]  # Includes ChromaDB and ML dependencies

Basic E2E Pipeline

from agentic_sdlc import AgentBridge
from pathlib import Path

# Initialize bridge with preferred LLM
bridge = AgentBridge(
    project_dir=Path("."),
    llm_provider="gemini",
    llm_model="gemini-2.0-flash"
)

# Run full E2E pipeline
response = bridge.process_request_enhanced("Implement JWT auth in FastAPI")

print(f"Detected Domain: {response.metadata['domain']}")
print(f"Optimized Prompt: {response.skill_instructions[:100]}...")

---

🏗 Architecture

agentic_sdlc/
├── bridge/          # Primary E2E Pipeline (AgentBridge)
├── swarm/           # Swarm Intelligence (Orchestrator, MessageBus, Agents)
├── knowledge/       # RAG Layer (KnowledgeBase, VectorStore, Embeddings)
├── core/            # Config, Domain Engine, LLM Routing, Artifacts
├── prompts/         # Prompt Lab, Generator, Optimization
├── intelligence/    # Self-Learning, Reasoner
├── skills/          # Skill Registry & Generation
└── sdlc/            # Board & Task Tracking

---

📚 Documentation

• Quick Start Guide - Get up and running in 5 minutes.
• Architecture Deep-Dive - How the swarm and RAG layers work.
• LLM Setup - Configuring Gemini, OpenAI, and local Ollama.
• Context for Agents - For AI agents like Cursor or Antigravity to understand this framework.

---

🔌 Supported LLM Providers

Provider	Model	Tier
Google Gemini	gemini-2.0-flash	Free/Paid
OpenAI	gpt-4o, gpt-4o-mini	Paid
Anthropic	claude-3-5-sonnet	Paid
Ollama	llama3.2, mistral	FREE / Local

---

🧪 Testing

pytest tests/unit/swarm  # Test swarm multi-agent logic
pytest tests/unit/knowledge # Test RAG retrieval

---

📄 License

MIT License. Developed by Dao Quang Truong | GitHub