Build an MCP Server in Python | Tools, Resources & Client (A03 Updated OCT26)

Build an MCP server in Python, connect a real client, and inspect the messages behind the result. Advanced A03 covers tools, resources, prompts, strict inputs, tool errors, protocol errors, and current versus legacy protocol behavior. This is a fictional, local, read-only training lab using the pinned MCP Python SDK v2.3.0 on Linux with Python 3.12.14. It tests six training-kit records, ten ordinary cases in each of two protocol modes, fourteen local loader cases and six additional failure/mode controls. The included independent checker compares recorded messages and results with the source bytes. No LLM inference, remote authentication, government tenant integration or organizational authorization is established by this lab. This advanced series runs alongside the original AI for GIs course. Original detailed Class 18:    • Module 4 Class 18 - Building AI Agents | A...   Corrected companion Class 18:    • AI Agents & MCP Explained: Palantir AIP + ...   Prerequisite: Advanced A02, Harnesses Models and Working Sessions:    • AI Agent Harness Explained: Codex, Claude ...   The course package includes an editable slide deck, student reference, runnable lab and dependency lock. Exact source locations and produced-media checks are retained in the private Citation Integrity Document. Public primary references: https://modelcontextprotocol.io/speci... and https://py.sdk.modelcontextprotocol.i... Stay connected: YouTube:    / @webb8020   X: https://x.com/webb8020 TikTok:   / justin.webb   Instagram:   / pstrategus   LinkedIn:   / tjustinwebb   Join us on NIPR Teams: AI for GIs. CHAPTERS 00:00 Build a local MCP server and client 01:40 Keep the exercise inside its boundary 04:22 Protocol, SDK and Python versions 06:59 Design the access contract 09:31 Read a resource through its address 11:16 Prompt templates do not execute tools 12:54 Keep diagnostics off protocol stdout 15:42 Tool errors and protocol errors 17:30 Code enforces the boundary 18:27 A version pin is an assertion 20:16 Reproduce and check the lab 21:56 Continue the advanced series AI-assisted course preparation and approved synthetic presenter/voice. #MCP #Python #AIforGIs