Beyond the Prompt: Leveraging MCP and PowerShell for Custom LLM Actions
summary
Builds a working MCP server over stdio in PowerShell that lists tools, accepts JSON-RPC messages, and runs concrete commands like get-weather and get-open-connections; shows both stdio and HTTP transports, logs protocol JSON to files, and demos successful calls from LM Studio and Claude as well as a local LLM with ~3B params. Provides an installable PowerShell SDK (MCP.SDK), enforces JSON schema guardrails, and confirms the server actually executes and returns results.
AI-written summary · about
chapters (16)
- 0:00MCP and PowerShell integration
- 1:12Copy-paste loop with AI tools
- 3:18MCP as USB-C for LLMs
- 5:50MCP components and workflow
- 10:49Live MCP demo
- 18:46MCP servers as JSON over stdin/stdout
- 26:40Basic MCP server in PowerShell
- 32:20Adding a second tool to server
- 39:49Context window management
- 45:16Designing agent-friendly tools
- 50:09MCP server deployment options
- 57:55MCP SDK for PowerShell
- 1:01:29Advanced SDK features
- 1:12:24Key takeaways and use cases
- 1:17:25Testing MCP servers
- 1:20:18Q&A
related talks
43mLLMs, Meet Your Tools: The Magic of Model Context Protocol
34mLLMs, Meet Your Tools: The Magic of Model Context Protocol
24mUnderstanding the Architecture of MCP (Model Context Protocol)
44mSupercharging DevOps with MCP (Without Opening a Security Hole)
01h 01m.NET Meets MCP: Build Your Own AI-Powered Service with C#
46mMCP for LLM Agents, Infrastructure APIs & Data Services
35mMCP, or not MCP
35mMCP in Action: Real-World Use Cases & Design Patterns