reasonflow 0.1.0

A workflow orchestration framework for LLM applications

ReasonFlow

A powerful workflow orchestration framework for LLM applications with built-in RAG support, agent management, and multi-modal processing capabilities.

Features

Core Components

• Workflow Orchestration

  • DAG-based workflow execution
  • Dynamic task scheduling with priority management
  • State persistence and versioning
  • Real-time status tracking
  • Error handling and retry mechanisms

• Agent System

  • Multiple LLM providers (OpenAI, Groq, Ollama, Anthropic)
  • RAG integration with vector databases
  • Custom task execution
  • API connector agents
  • Dynamic agent creation and configuration

• Document Management

  • PDF processing and text extraction
  • Vector database integration (FAISS)
  • Document versioning
  • Full-text search capabilities
  • Multiple storage backends (S3, MinIO)

Integrations

• LLM Providers

  • OpenAI (gpt-4o, gpt-4o-mini)
  • Groq (gemma, llama)
  • Ollama (local models)
  • Anthropic (claude)
  • ReasonChain

• Vector Databases

  • FAISS (CPU/GPU)
  • Pinecone
  • Milvus
  • Qdrant
  • Weaviate

• Storage Solutions

  • Firebase
  • AWS S3
  • MinIO
  • Local storage

Installation

Basic Installation (CPU)

pip install reasonflow

GPU Support (Recommended for large-scale deployments)

# Create conda environment
conda create -n reasonflow python=3.10
conda activate reasonflow

# Install FAISS GPU
conda install -c conda-forge faiss-gpu

# Install ReasonFlow
pip install reasonflow

Development Installation

git clone https://github.com/yourusername/reasonflow.git
cd reasonflow
pip install -e ".[dev]"

Quick Start

1. Configure Environment

Create a .env file:

OPENAI_API_KEY=your_openai_key
GROQ_API_KEY=your_groq_key
VECTOR_DB_PATH=vector_dbs/

2. Create a Simple Workflow

from reasonflow.orchestrator.workflow_builder import WorkflowBuilder
from reasonflow.integrations.llm_integrations import LLMIntegration
import os

# Initialize components
workflow_builder = WorkflowBuilder()

# Create LLM agent
llm = LLMIntegration(
    provider="openai",
    model="gpt-4o",
    api_key=os.getenv("OPENAI_API_KEY")
)

# Define workflow
workflow_config = {
    "tasks": {
        "analyze": {
            "type": "llm",
            "config": {
                "agent": llm,
                "params": {"prompt": "Analyze this text"}
            }
        }
    }
}

# Create and execute workflow
workflow_id = workflow_builder.create_workflow(workflow_config)
results = workflow_builder.execute_workflow(workflow_id)

3. RAG Integration Example

from reasonflow.integrations.rag_integrations import RAGIntegration

# Initialize RAG
rag = RAGIntegration(
    db_path="my_vectors.index",
    db_type="faiss",
    embedding_model="all-MiniLM-L6-v2"
)

# Add documents
rag.add_documents("document.pdf")

# Search
results = rag.search("What is the revenue growth?", top_k=5)

Project Structure

reasonflow/
├── agents/                 # Agent implementations
├── config/                 # Configuration files
├── integrations/          # External service integrations
├── orchestrator/          # Workflow management
├── persistence/           # State and data persistence
├── sdk/                   # Python and JavaScript SDKs
└── tasks/                 # Task management

Development

Running Tests

pytest tests/

Building Package

python build.py

Code Style

black reasonflow/
isort reasonflow/
flake8 reasonflow/

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

MIT License - see LICENSE for details.

Support

• Documentation: docs.reasonflow.ai
• Issues: GitHub Issues
• Discord: Join our community