Skip to main content

Capstone — Autonomous Research Agent

Objective​

Build a multi-agent system using LangGraph that can research a complex topic, execute code to gather data, and write a comprehensive report.

Architecture​

  • Planner Agent: Breaks the prompt into sub-tasks.
  • Researcher Agent: Uses a web search tool to find information.
  • Coder Agent: Writes and executes Python code in a sandbox to process data.
  • Summarizer Agent: Compiles findings into a Markdown report.

Requirements​

  1. Implement state management using LangGraph.
  2. Include a "Human-in-the-Loop" checkpoint before the final report is generated.
  3. Ensure the Coder agent is sandboxed securely.

Deliverables​

  • Source code in a GitHub repository.
  • A sample generated report on "The impact of quantum computing on RSA encryption".