Production patterns & what's Next

You've built MCP servers and clients from scratch. Let's do a speed round to lock in the concepts, then look at where to go from here.

Production patterns to keep in mind: - Use Streamable HTTP transport for deployed servers - Implement error handling per-server in multi-server setups - Use AsyncExitStack to manage concurrent server connections - Consider connection pooling for high-traffic servers - Log tool calls with server attribution for debugging

Where to go from here: - Explore the MCP server registry for pre-built servers (filesystem, GitHub, databases) - Build your own servers for internal tools and APIs - Add MCP to existing AI applications - Check out the AI Agent Design Patterns course to combine MCP with advanced agent architectures