🚀 AgenticaForge: The Full-Stack Agentic OS

AgenticaForge is a production-grade AI agent orchestration platform that bridges visual intuition with professional-grade backend logic. It integrates the AgentForge Visual Dashboard with the Agentica (agentic-flow) Engine, powered by a high-performance Node.js service architecture.

🏗️ System Architecture

AgenticaForge is built on a distributed, event-driven architecture designed for high-concurrency agentic swarms.

• Frontend (AgentForge): A React 19 visual dashboard for designing agent topologies, inspecting memory tiers (Hot/Warm/Cold), and monitoring real-time telemetry from the SONA training engine.
• Backend API (Server): An Express service that manages the persistence layer (Postgres/Drizzle), exposes agent controllers, and handles the "Simulation to Reality" bridge.
• Background Worker: A dedicated task processor that executes LLM calls, handles long-running agentic loops, and generates vector embeddings for the ReasoningBank.
• Core Engine (Agentica): The agentic-flow toolkit—a production-ready library featuring 66 specialized agents, 213 MCP tools, and quantum-resistant version control.

✨ Core Features

🧩 Visual Swarm Orchestration

Design complex multi-agent systems using a visual graph editor (React Flow). Transition seamlessly from Mock Simulations in the browser to Real Swarm Execution on the backend with a single click.

🧠 ReasoningBank (Three-Tier Memory)

A sophisticated memory management system inspired by neurological models:

• Hot Memory: Real-time context, rapid access, high decay.
• Warm Memory: Recent patterns and frequently accessed clusters.
• Cold Memory: Long-term persistent knowledge, indexed via HNSW for semantic retrieval.

🌐 Multi-Provider Intelligence

Native support for OpenAI (O1/GPT-4o), Anthropic (Claude 3.5/Sonnet), HuggingFace, and Ollama. The ModelRouter automatically handles fallback, load balancing, and provider-specific optimizations.

📊 SONA Neural Telemetry

Monitor the health and "learning" of your agents in real-time. High-resolution charts track reward curves, pattern density, and loss metrics as agents refine their strategies.

---

🚀 Quick Start

AgenticaForge is a monorepo that manages its own dependencies and service lifecycle.

1. Clone & Setup

git clone https://github.com/QRcode1337/AgenticaForge.git
cd AgenticaForge
npm install

2. Configure Environment

Update the .env files in both the server/ and worker/ directories with your API keys:

OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
SUPABASE_URL=...
DATABASE_URL=...

3. Launch the Stack

Start the development server, background worker, and frontend dashboard concurrently:

npm run dev:all

• Dashboard: http://localhost:3000
• API Server: http://localhost:3001
• Worker: Running in background via tsx

---

🛠️ Tech Stack

• Frontend: React 19, Vite 7, Tailwind 4, React Flow, Three.js, Recharts.
• Backend: Node.js (v25+), Express, Drizzle ORM, Postgres.
• Engine: Agentica (agentic-flow), AgentDB, SONA Neural Layer.
• Runtime: tsx (TypeScript Execution Engine), Concurrently.

📁 Project Structure

AgenticaForge/
├── agentforge/      # Next.js / React Visual Interface
├── agentica/        # Core Agentic Engine (agentic-flow SDK)
├── server/          # Express API Gateway & Shared DB Layer
├── worker/          # Background LLM Executor & Embedding Service
├── conductor/       # Orchestration plans and integration logic
├── docs/            # Architecture & Getting Started guides
└── package.json     # Root dev:all script and monorepo config

📜 License

MIT © QRcode1337