An open API service indexing awesome lists of open source software.

https://github.com/divakarkumarp/langgraph_agent

LangGraph Agent Implementation - RAG, Multi, Core Agents
https://github.com/divakarkumarp/langgraph_agent

agentic-ai agents langchain langgraph multiagent rag

Last synced: 3 months ago
JSON representation

LangGraph Agent Implementation - RAG, Multi, Core Agents

Awesome Lists containing this project

README

          

# LangGraph Agent Implementation Project

## Overview

This project showcases different implementations of LangGraph-based agents and RAG (Retrieval Augmented Generation) tools, demonstrating various approaches to building intelligent conversational systems.

## Agent Implementations

### RAG Agents

| Type | Description | Category |
|------|-------------|----------|
| 🤖 Self RAG | [Enhanced RAG with adaptive retrieval](LangGraph_1/types_RAG-with-LangGraph/self_rag.ipynb) | 📚 RAG |
| 🔄 Corrective RAG | [Improved accuracy with correction mechanisms](LangGraph_1/types_RAG-with-LangGraph/corrective_rag.ipynb) | 📚 RAG |
| 💾 SQL Agent | [Database interactions and processing](LangGraph_1/types_RAG-with-LangGraph/sql_agent.ipynb) | 🗃️ Database |
| 🎯 Agentic RAG | [Advanced RAG capabilities](LangGraph_1/types_RAG-with-LangGraph/Agentic_rag.ipynb) | 📚 RAG |
| 🎫 Customer Support | [Autonomous support system](LangGraph_1/types_RAG-with-LangGraph/autonomous_Customer_support_agent.ipynb) | 💁 Support |

### Multi-Agent Systems

| Type | Description | Category |
|------|-------------|----------|
| 👥 Multi-Agent System | [Coordinated supervisor system](LangGraph_1/Multi-Agent-Systems/multi_agent_supervisor.ipynb) | 🤝 Multi-agent |
| 📊 Research Analyst | [Research and analysis system](LangGraph_1/Multi-Agent-Systems/multiagent-research_analyst.ipynb) | 🤝 Multi-agent |
| 🌐 Network Agent | [Networked agent system](LangGraph_1/Multi-Agent-Systems/network_multiagent_system.ipynb) | 🤝 Multi-agent |
| 👨‍💼 Supervisor Agent | [Core supervisor implementation](LangGraph_1/Multi-Agent-Systems/supervisor_agent.ipynb) | 🤝 Multi-agent |

### Core Agents

| Type | Description | Category |
|------|-------------|----------|
| 🤔 ReAct Agent | [ReAct pattern implementation](ReAct_Agent.ipynb) | ⚡ Core |
| 🔧 Tool Agent | [Tool-based agent implementation](tool_agent.ipynb) | ⚡ Core |

## Project Structure

```plaintext
.
├── LangGraph_1/
│ ├── types_RAG-with-LangGraph/
│ │ ├── self_rag.ipynb
│ │ ├── corrective_rag.ipynb
│ │ ├── sql_agent.ipynb
│ │ ├── Agentic_rag.ipynb
│ │ └── autonomous_Customer_support_agent.ipynb
│ ├── Multi-Agent-Systems/
│ │ ├── multi_agent_supervisor.ipynb
│ │ ├── multiagent-research_analyst.ipynb
│ │ ├── multiagent.ipynb
│ │ ├── network_multiagent_system.ipynb
│ │ ├── network.ipynb
│ │ └── supervisor_agent.ipynb
│ └── human-in-loop.ipynb
├── ReAct_Agent.ipynb
├── tool_agent.ipynb
└── requirements.txt
```

## Features

### 1. RAG Implementations

- **Self-RAG**: Enhanced retrieval with self-reflection
- **Corrective RAG**: Real-time correction mechanisms
- **SQL RAG**: Database interaction capabilities
- **Agentic RAG**: Advanced agent-based RAG
- **Customer Support RAG**: Knowledge base integration

### 2. Multi-Agent Systems

- Supervisor-based architecture
- Research and analysis capabilities
- Networked agent communication
- Task delegation and management

### 3. Core Implementations

- ReAct pattern implementation
- Tool-based agent systems
- Human-in-the-loop processing

## Technology Stack

- **LangGraph**: Core framework for agent workflows
- **LangChain**: LLM integration and tools
- **Jupyter**: Interactive notebook environment
- **SQLite**: Database for SQL agent demonstrations

## Getting Started

1. Install dependencies:

```bash
pip install -r requirements.txt
```

2. Set up environment variables:

```bash
# Create .env file with your API keys
OPENAI_API_KEY=your_key
LANGCHAIN_API_KEY=your_key
```

3. Run the Jupyter notebooks:

```bash
jupyter notebook
```

## Development Best Practices

### 1. Agent Design

- Clear node separation
- Proper error handling
- Modular design patterns

### 2. RAG Implementation

- Optimize chunk sizes
- Configure retrieval parameters
- Implement validation

### 3. State Management

- Type-safe transitions
- Proper validation
- Error handling

## Contributing

Contributions welcome for:

- New agent implementations
- RAG enhancements
- Documentation improvements
- Bug fixes

## License

This project is licensed under the MIT License - see the LICENSE file for details.

## Agent Workflow Architectures

### RAG Agent Workflows

#### Self RAG Workflow

```mermaid
graph TB
A[Start] --> B[Query Input]
B --> C[Generate Initial Response]
C --> D[Self-Reflection]
D --> E[RAG Retrieval]
E --> F[Response Refinement]
F --> G[Final Response]
```

#### Corrective RAG Workflow

```mermaid
graph TB
A[Start] --> B[Query Input]
B --> C[Initial RAG]
C --> D[Generate Response]
D --> E[Error Detection]
E --> F{Has Errors?}
F -->|Yes| G[Correction]
G --> C
F -->|No| H[Final Response]
```

#### SQL Agent Workflow

```mermaid
graph TB
A[Start] --> B[Natural Language Query]
B --> C[Query Analysis]
C --> D[SQL Generation]
D --> E[Database Query]
E --> F[Result Processing]
F --> G[Natural Language Response]
```

#### Agentic RAG Workflow

```mermaid
graph TB
A[Start] --> B[Query Processing]
B --> C[Task Planning]
C --> D[RAG Retrieval]
D --> E[Context Integration]
E --> F[Response Generation]
F --> G[Final Response]
```

### Multi-Agent System Workflows

#### Supervisor Agent Workflow

```mermaid
graph TB
A[Start] --> B[Task Distribution]
B --> C[Agent Assignment]
C --> D[Task Execution]
D --> E[Progress Monitoring]
E --> F{Task Complete?}
F -->|No| D
F -->|Yes| G[Result Aggregation]
G --> H[Final Output]
```

#### Research Analyst Workflow

```mermaid
graph TB
A[Start] --> B[Research Query]
B --> C[Source Collection]
C --> D[Information Analysis]
D --> E[Data Synthesis]
E --> F[Report Generation]
F --> G[Review & Validation]
G --> H[Final Report]
```

#### Network Agent Workflow

```mermaid
graph TB
A[Start] --> B[Task Reception]
B --> C[Agent Network Distribution]
C --> D[Parallel Processing]
D --> E[Result Collection]
E --> F[Response Synthesis]
F --> G[Final Output]
```

### Core Agent Workflows

#### ReAct Agent Workflow

```mermaid
graph TB
A[Start] --> B[Task Input]
B --> C[Thought Generation]
C --> D[Action Selection]
D --> E[Action Execution]
E --> F[Observation]
F --> G{Task Complete?}
G -->|No| C
G -->|Yes| H[Final Output]
```

#### Tool Agent Workflow

```mermaid
graph TB
A[Start] --> B[Task Analysis]
B --> C[Tool Selection]
C --> D[Tool Execution]
D --> E[Result Processing]
E --> F{More Tools Needed?}
F -->|Yes| C
F -->|No| G[Response Generation]
G --> H[End]
```