Multi-Agent Task Execution System (LangGraph)

Overview

This project implements a controlled multi-agent workflow using LangGraph and LLMs to execute complex tasks in a structured and reliable manner.

Instead of relying on a single LLM response, the system follows an iterative loop: Planning → Execution → Evaluation → Retry/Advance

This approach improves robustness, traceability, and control over task completion.

---

Architecture

The system is composed of three core agents:

Planner

• Decomposes a high-level task into sequential subtasks
• Produces a structured list of actionable steps

Executor

• Executes each subtask step-by-step
• Uses previous results as context for continuity

Critic

• Evaluates the output of each step
• Returns:
  • PASS → proceed to next step
  • FAIL → retry (with limit)

Key Features

• Iterative multi-agent loop (Plan → Execute → Evaluate)
• Controlled termination (prevents infinite loops)
• Retry mechanism with limits
• Self-evaluation using a critic agent
• Structured state management via LangGraph
• JSON output logging for reproducibility
• Secure API key handling using environment variables

---

Tech Stack

• Python
• LangGraph
• LangChain
• OpenAI API

---

Installation

Clone the repository:

git clone https://github.com/your-username/repo-name.git
cd repo-name

Install dependencies

pip install -r requirements.txt

How to Run

python multi_agent_demo.py