> Markdown version of [/videos/1727-build-a-multi-agent-role-playing-game-master-with-strands-agents?t=165](https://www.wearedevelopers.com/videos/1727-build-a-multi-agent-role-playing-game-master-with-strands-agents?t=165). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Build a Multi-Agent Role-Playing Game Master with Strands Agents Standardizing agent-to-tool communication eliminates redundant code. Learn how to architect a scalable, multi-agent RPG Game Master using Model Context Protocol and the open-source Strands SDK. - **Speakers:** [Chris Heilmann](https://www.wearedevelopers.com/@chris-heilmann), [Daniel Cranney](https://www.wearedevelopers.com/@daniel-cranney), Tiffany Souterre - **Event:** WeAreDevelopers LIVE - **Published:** September 30, 2025 - **Duration:** 41:33 - **URL:** https://www.wearedevelopers.com/videos/1727-build-a-multi-agent-role-playing-game-master-with-strands-agents ## Summary Transitioning from traditional large language models to AI agents effectively gives standard models the functional tools needed to execute actions via external APIs. However, managing connections between diverse agents and isolated tools often leads to redundant code and complex integrations. The Model Context Protocol (MCP) solves this by standardizing agent-to-tool communication, effectively functioning as APIs reinvented for agents. Running parallel to this is the Agent-to-Agent (A2A) protocol, which streamlines how autonomous models discover and collaborate with each other inside complex workflows.<br><br>To demonstrate these protocols in practice, an AI-driven role-playing Game Master can be built using the open-source Strands Agents SDK. Strands provides a model-agnostic approach, allowing developers to seamlessly switch inference from cloud-based frameworks like Claude 4 on AWS Bedrock to local instances like Llama 3 via Ollama. By equipping a base agent with custom operations—such as a specialized dice-rolling function mounted onto localized MCP servers—developers can empower any connected network to dynamically pull tailored capabilities.<br><br>Scaling multi-agent orchestration ultimately shifts developer focus from writing exhaustive imperative logic to crafting precise system prompts. In the custom Game Master architecture, specialized secondary agents (like an NPC character creator querying isolated databases) run on independent A2A servers. When prompted, the primary agent fetches descriptive agent cards to understand available network nodes, autonomously deciding whether to query the MCP for a dice roll or ping the specialized agent to generate character stats. This architecture proves that decoupling functional execution from conversational memory is critical for building robust, scalable AI ecosystems. **Keywords:** model context protocol, agent-to-agent protocol, multi-agent orchestration, strands agents sdk, custom mcp servers, ai tool integration, local llm inference, aws bedrock deployment, ollama containerization, system prompt engineering, dynamic tool discovery, agent discovery cards, autonomous ai agents, llm action networks, rpg game master architecture ## Chapters 1. **Defining autonomous agents and tool integrations** (00:02) — Adding external tools to conversational AI converts static text generators into actionable software entities. 1. **Standardizing agent communication with MCP and A2A** (02:45) — The Model Context Protocol and Agent-to-Agent protocol standardize connections between models, tools, and collaborating agents. 1. **Initializing a simple agent using the Strands SDK** (05:17) — Installing the Strands framework and instantiating an agent enables rapid connection to cloud-based language models. 1. **Executing query resolution with local Ollama models** (08:56) — Connecting the framework to a locally hosted model provides offline capabilities for localized testing. 1. **Importing built-in functions to overcome context limitations** (11:27) — Applying pre-built community tools to an agent provides real-time information access previously missing from training data. 1. **Writing and executing files dynamically with AI tools** (14:16) — Adding file manipulation capabilities lets models autonomously author, read, and execute external scripts directly. 1. **Creating custom agent tools using Python decorators** (16:36) — Wrapping a standalone Python function with a configuration decorator exposes internal application logic to model orchestration. 1. **Deploying custom functions onto an MCP server** (20:39) — Migrating native tools to a dedicated server endpoint enables standardized function consumption across disparate models. 1. **Building specialized sub-agents with the A2A server** (25:47) — Configuring an independent agent with focused prompts allows the system to delegate highly specific computational workloads. 1. **Connecting multi-agent systems for workflow orchestration** (31:43) — Passing endpoint parameters back to a master orchestrator empowers language models to collaboratively request tasks and exchange data. 1. **Testing the completed multi-agent database workflow** (37:23) — Deploying the fully assembled architecture showcases dynamic interactions spanning interface inputs, local APIs, and specialized knowledge bases. ## Related Moments - [Using large language models as tabletop game masters](https://www.wearedevelopers.com/videos/1725-wearedevelopers-live-build-a-multi-ai-agents-game-master-with-strands-our-weekly-web-finds) (from "WeAreDevelopers LIVE - Build a multi AI agents game master with Strands & our weekly web finds") - [Understanding foundational concepts of LLMs, agents, and MCPs](https://www.wearedevelopers.com/videos/1768-boost-productivity-with-ai-figma-playwright-mcp-workflows-aris-markogiannakis) (from "Boost Productivity with AI: Figma & Playwright MCP Workflows - Aris Markogiannakis") - [Standardizing autonomous agent logic with model context protocol](https://www.wearedevelopers.com/videos/1554-java-meets-ai-empowering-spring-developers-to-build-intelligent-apps) (from "Java Meets AI: Empowering Spring Developers to Build Intelligent Apps") - [Architectural patterns for composing dynamic AI agents](https://www.wearedevelopers.com/videos/1629-beyond-chatbots-how-to-build-agentic-ai-systems) (from "Beyond Chatbots: How to build Agentic AI systems") - [Moving from playing with artificial intelligence to business implementation](https://www.wearedevelopers.com/videos/2136-moving-from-playing-with-ai-to-implementing-ai-lee-stott) (from "Moving from Playing with AI to Implementing AI - Lee Stott") - [Building autonomous functions with conversational agent frameworks](https://www.wearedevelopers.com/videos/1624-30-powerful-aws-hacks-in-just-30-minutes-boost-your-developer-productivity) (from "30 powerful AWS hacks in just 30 minutes: Boost your developer productivity") ## Related Articles - [From Prototype to Production: Build AI Agents with This Free 4-Course Learning Path](https://www.wearedevelopers.com/magazine/655-from-prototype-to-production-build-ai-agents-with-this-free-4-course-learning-path) - [Everything a Developer Needs to Know About MCP with Neo4j](https://www.wearedevelopers.com/magazine/604-everything-a-developer-needs-to-know-about-mcp-with-neo4j) - [A 5-Step Open-Source Setup for Agentic Engineering](https://www.wearedevelopers.com/magazine/738-a-5-step-open-source-setup-for-agentic-engineering) - [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care) ## Related Jobs - [Senior Backend Developer — AI: MCP & Agent Engine](https://www.wearedevelopers.com/jobs/48297-senior-backend-developer-ai-mcp-agent-engine) at **basebox GmbH** - [Principal Product Manager, Agent Platform](https://www.wearedevelopers.com/jobs/ext/277541-principal-product-manager-agent-platform) at **GitHub** - [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace** - [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat** - [Principal Field Architect - AI Agents](https://www.wearedevelopers.com/jobs/ext/1442858-principal-field-architect-ai-agents) at **Twilio** - [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**