This MCP (Model Context Protocol) example shows how to implement a Model Context Protocol server within the watsonx.ai Flows Engine. It demonstrates how an agent can invoke tools while maintaining a consistent, shared model context across tool calls. The MCP server coordinates context propagation, enabling tools to read and update the shared state as the agent progresses through a workflow. It leverages the Flows Engine's tool-calling capabilities and serves as a reference for building end-to-end AI tool workflows that require cross-tool context management. The example is located under examples/mcp in the repository and complements the broader demonstrations of tool integrations and patterns described in the README, including LangGraph, LangChain, watsonx.ai, and OpenAI tool-calling scenarios.