Cheatsheet AI

Cheatsheet AI

Summary: AI skills, agents, etc.
Date: 29 May 2026
Last update: 9 juli 2026

Note: As AI is shifting in an enormous speed, this page may not be up to date.

General

Wiki AI Statement

Note that all pages on this wiki are human made. AI could have been used in assisting creating components of pages and or public github resources. However, everything is tested by me, and after some tests in writing the actual pages with AI, I concluded that I could do better, meaning that the pages themselves are, except for the ones I tested with, written by me. In the end it comes down to:

If I can't write about it, I don't understand it.

AI and Your Mind

AI can blindfold you, building up trust, and then betray you. Treat AI like an intern and check everything. To keep sharp, allow yourself some AI-free time and space to boost your own critical thinking and creativity.

Costs

Tokens

When using AI, tokens are the currency. See Tech in Five - Tokens & Tokenomics from John Savill's Technical Training Youtube channel for an excellent explaination.

Budgets

For Co-Pilot, set an additional budget on top of your plan here. Note that these are personal account budgets. See here to check wat happens if you exceed your credits.

Skills

For an overview of useful skills I've created check my public github.

Usefull Skills

Skills I'm currently testing

Community

Skill Specs

Skills must adhere to certain specifications

MCP Servers

Model Context Protocol (MCP) is a protocol that allows you to extend the capabilities of GitHub Copilot by integrating it with other systems. See Configure MCP servers for your repository for some examples and specifics.

For now, I use a mcp config file like this, located in the repository like this: .vscode/mcp.json

{
  "servers": {
    "drawio": {
      "type": "http",
      "url": "https://mcp.draw.io/mcp"
    },
    "azure-mcp-server": {
        "command": "npx",
        "args": [
          "-y",
          "@azure/mcp@latest",
          "server",
          "start"
        ]
    }
  },
  "inputs": []
}

For the latest version of my mcp.json, see mcp.json

Note that the drawio example is using a remote mcp server, while the Azure MCP server gets installed locally.

Note: there are various ways to install MCP servers. What works best depends on the specific MCP server. To use the npx installation method, you'll need node -v and npm -v

When you look into the file, you'll notice a small Start button to enable the server:

Start MCP server

You can start from there, or by using the command palette (ctrl+shift+p) → MCP: List Servers → Select and Start the server you want to start.

Draw.IO MCP Server

Source: Draw.io MCP Server

Besides installing the MCP server, also install the following VS Code resources:

If you want to work locally in VS Code with the created diagrams, make sure to install the extension “hediet.vscode-drawio”
Also install the skill drawio-mcp-diagramming

When using these combined resources, you can do a prompt for example like this:

Analyze the IaC in this repository and create an Azure resources and network drawio diagram. Make sure to only list resources that are explicitly created using the IaC files. If possible, list CIDRs and explicitly assigned IP addresses. If possible, add the name of the resources. Group the resources on both resource group as well as the network. Where possible, respect hub-spoke architecture in the diagram. Make sure to use Azure Icons as defined in the drawio-mcp-diagramming skill. Save the diagram as networkdiagram.drawio.png in the docs/diagrams folder.

Analyze the IaC in this repository and create an Azure Management drawio diagram. If possible, add Management Groups, Resource Groups, Policies, Permissions. Make sure to use Azure Icons as defined in the drawio-mcp-diagramming skill. Save the diagram as highleveloverview.drawio.png in the docs/diagrams folder.

Azure MCP Server

Source: Azure MCP Server