Welcome to FluxGraph Documentation

FluxGraph is an enterprise-grade AI agent orchestration framework for building intelligent, multi-agent systems with advanced memory, caching, and workflow capabilities.

Version 3.0.0 License MIT

Quick Start

Install FluxGraph:

pip install fluxgraph

Create your first agent:

from fluxgraph import FluxApp

app = FluxApp(title="My AI Agent", version="1.0.0")

@app.agent(name="greeter")
async def greeter(message: str):
    return {"response": f"Hello! You said: {message}"}

if __name__ == "__main__":
    app.run()

Features

🚀 Multi-Agent Orchestration
  • Build complex agent workflows

  • Dynamic agent chaining

  • Parallel execution support

🧠 Advanced Memory System
  • Episodic and semantic memory

  • PostgreSQL-backed persistence

  • Context-aware recall

High Performance
  • Built-in caching layer

  • Async-first architecture

  • Real-time analytics dashboard

🔌 LLM Provider Support
  • OpenAI, Anthropic, Google Gemini

  • Groq, Ollama, HuggingFace

  • Custom provider integration

🛠️ Production Ready
  • CORS support

  • Security middleware

  • Performance monitoring

  • Error handling & logging

Architecture Overview

FluxGraph follows a modular architecture:

┌─────────────────────────────────────┐
│         FastAPI Application         │
├─────────────────────────────────────┤
│      Agent Orchestrator Layer       │
├──────────┬──────────┬───────────────┤
│  Memory  │  Cache   │   Workflows   │
├──────────┴──────────┴───────────────┤
│        LLM Provider Adapters        │
├─────────────────────────────────────┤
│      Database & External APIs       │
└─────────────────────────────────────┘

Use Cases

Customer Support Agents

Intelligent chatbots with context retention and escalation handling

Lead Generation Systems

Automated sales qualification and CRM integration

Data Processing Pipelines

Multi-step workflows with AI-powered decision making

Knowledge Management

RAG-enabled question answering systems

Community

Indices and tables