📦 Agentic SDLC Kit

The Intelligence Layer for your Software Development Lifecycle. Installable, scalable, and self-learning.

Agentic SDLC is a portable AI-powered development framework that transforms any repository into an intelligent development ecosystem. It provides specialized AI experts, automated workflows, and a "Brain" that learns from every line of code you write.

---

🚀 Quick Start

1. Install the Kit (via Bun or Pip)

# Recommended
bun install agentic-sdlc

# Alternative
pip install agentic-sdlc

2. Initialize your Project

Navigate to your project root and run:

agentic init

This scaffolds the following in your project:

• .agent/ - AI Expert roles, rules, and workflows.
• docs/ - Project documentation and SDLC artifacts.
• agentic.yaml - Runtime configuration.

3. Start your first Sprint

agentic sprint start 1

---

🧠 What's inside the Kit?

👥 18 Specialized AI Roles

Assign tasks to specific experts directly in your IDE:

• @PM (Project Manager) - Planning & Tracking
• @SA (System Architect) - Design & Architecture
• @DEV (Developer) - Implementation
• @TESTER (QA) - Verification & Self-Healing
• @SECA (Security) - Audits & Safety
• @RESEARCH (Specialist) - Technical Research & Swarms

⚡ 23 Automated Workflows

Execute complete SDLC phases with simple slash commands:

• /cycle - Research → Plan → Code → Review in one go.
• /orchestrator - Full automation of complex features.
• /swarm - Intelligent multi-agent routing.
• /concurrent - Execute multiple roles (SA, UIUX, PO) in parallel.
• /synthesize - Mixture of Agents (MoA) synthesis.
• /heal - Automated bug detection and auto-fixing.
• /ab - Generate and compare architectural alternatives.

3-Layer Architecture

The system follows a concentric design ensuring safety and consistency:

• Layer 1: Core - GEMINI.md, Rules, and Workflows.
• Layer 2: Intelligence - 26 Sub-Agents (Brain, SwarmRouter, Self-Learning).
• Layer 3: Infrastructure - CLI, SDK, AOP (Agent Orchestration Protocol).

---

🛠️ Usage

CLI Commands

The kit provides a unified entry point:

agentic status              # View current SDLC state
agentic init-state --sprint 1  # Initialize brain state for a sprint
agentic heal --code src/main.py # Run self-healing on a file
agentic gate list           # View pending human-in-the-loop approvals

IDE Integration (Slash Commands)

If you use an AI-powered IDE (Cursor, Windsurf, etc.), simply reference the workflows:

@DEV /cycle Implement user authentication
@PM /planning Create a plan for the next feature

---

🐍 SDK Usage (Python)

Integrate the Agentic Brain directly into your own scripts:

from agentic_sdlc import Learner, SprintManager, get_project_root

# Get the current project context
root = get_project_root()

# Record a learning event
learner = Learner()
learner.learn("Refactored database layer for performance")

# Manage sprints
sm = SprintManager()
sm.create_sprint("Feature Alpha", "Deliver MVP")

---

🏗️ Enterprise Features

• 🛡️ Sandboxing: Execute agent code in isolated Docker containers.
• 🩹 Self-Healing: Automated feedback loops that learn from test failures.
• 🌊 Swarms Orchestration: Universal routing, parallel execution, and expert synthesis.
• 📡 AOP Protocol: Distributed Agent Orchestration Protocol for distributed AI.
• Knowledge Graph: Optional Neo4j integration for cross-project intelligence.
• Local LLM Support: Full compatibility with Ollama for privacy-first development.

---

📄 License

MIT License. See LICENSE for details.

---

Developed by Dao Quang Truong | GitHub