Skip to content

Technical implementation of high-reasoning AI agents using the AWS Strands SDK, featuring autonomous tool-use, Model Context Protocol (MCP) integration, and multi-agent coordination.

License

Notifications You must be signed in to change notification settings

noecrn/mcp-agent-framework

Repository files navigation

🐉 Once Upon Agentic AI: The Strands Quest

AWS Workshop Python SDK

"Roll for Initiative... in Python!"

This repository contains my code and progress through the AWS Strands SDK Workshop. It tracks my transformation from a coding apprentice into a master of AI agent orchestration, building a digital adventuring party capable of autonomous thought and collaboration.

🗺️ The Adventure Map

The project is structured into five epic chapters, each building upon the last:

🧙‍♂️ Chapter 1: Agent Summoning

  • Goal: Master the fundamental ritual of agent creation.
  • Key Learning: Configuring system prompts and model parameters within the Strands framework.

⚔️ Chapter 2: The Adventurer's Arsenal

  • Goal: Equip agents with built-in magical tools.
  • Key Learning: Implementing autonomous web scraping and information gathering with tool safety mechanisms.

🔨 Chapter 3: The Art of Magical Forging

  • Goal: Create the legendary Dice of Destiny.
  • Key Learning: Transforming Python functions into agent-executable tools using the @tool decorator.

🌐 Chapter 4: Planar Portals (MCP)

  • Goal: Connect to external realms via Model Context Protocol.
  • Key Learning: Building MCP servers and clients to bridge agents with external services.

🏰 Chapter 5: The Grand Alliance (A2A)

  • Goal: Command a full D&D adventuring party.
  • Key Learning: Orchestrating Agent-to-Agent (A2A) communication for complex, distributed AI tasks.

🛠️ Tech Stack

  • Language: Python
  • Framework: Strands SDK
  • AI Models: Amazon Bedrock (Claude 3 / 3.5)
  • Integration: Model Context Protocol (MCP)

🚀 Installation & Setup

  1. Clone the Quest Log
git clone https://github.com/noecrn/mcp-agent-framework.git
cd strands-quest
  1. Prepare your Environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt
  1. Configure AWS Credentials Ensure your AWS CLI is configured with permissions for Amazon Bedrock.

🎲 Journey Highlights

  • The Dice of Destiny: A custom tool allowing agents to simulate high-stakes RPG rolls.
  • Multi-Agent Coordination: Watching a "Dungeon Master" agent delegate lore-checking to a "Librarian" agent.

“The best way to predict the future is to build the agents that will create it.”

About

Technical implementation of high-reasoning AI agents using the AWS Strands SDK, featuring autonomous tool-use, Model Context Protocol (MCP) integration, and multi-agent coordination.

Topics

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages