The art of simplicity

As the list of available tools keeps growing, sooner or later something will not work and some debugging becomes necessary. In this post I look into ways to troubleshoot your MCP integration. Let's dive in! To understand what we need to debug you need to be aware of the architecture of an MCP integration. It follows a client-server architecture , where AI models can request data through a host with an MCP client (e.g. GitHub Copilot, Claude Desktop, …) from MCP servers, which then retrieve relevant information from local or remote sources. This means that when a problem occurs that there (at least) 2 places to look at. Debugging the MCP client Let’s start by looking into the client. The way you need to debug the client is completely dependent on the host. I’ll focus on GitHub Copilot and Claude Desktop. GitHub Copilot When VSCode encounters an error while trying to interact with an MCP server, you get a red error indicator in the Chat window: Click on the icon and...

On Monday I talked about interacting with Word and Powerpoint documents directly through GitHub Copilot. After writing that post , a colleague reached out to me asking if the same is possible through Claude Desktop . As the original MCP protocol was created by Anthropic(the company that created Claude Desktop), the answer is 'of course'. Remark: If you want a general introduction about MCP servers, check out my previous post . The use case Part of my job is working on presales activities. Typically, this involves answering RFI's and RFP's by creating solution designs, project plans and describe the supporting processes. The end result is typically a Word document containing all the details about our offer in combination with a PowerPoint to present our offer to the customer. I was wondering if I could let an AI agent interact directly with these documents to help me create and finetune our proposals. Use an MCP server to talk to Microsoft Office To realize th...

Through the MCP integration in GitHub Copilot, your AI agent is no longer limited to interactions with your IDE but can interact with your local computer and the outside world. A problem is that this open up a new range of possible attack vectors and malicious actors. So be careful when downloading a random MCP server example from the Internet. GitHub Copilot Agents helps you by asking by default permission to execute a task: Only by clicking on Continue the MCP server instance is called, and the tool is executed. Of course, it can become annoying to confirm this over and over again. Therefore, you can choose between multiple options: Allow in this Session Allow in this Workspace Always allow This gives you full control to balance between security and convenience. If you really want to just auto-approve everything (not recommended) set this in your settings: chat.tools.autoApprove: true Happy (vibe) coding! More information Use MCP servers in VS Code (Prev...

Part of my job is working on presales activities. Typically, this involves answering RFI's and RFP's by creating solution designs, project plans and describe the supporting processes. The end result is typically a Word document containing all the details about our offer in combination with a PowerPoint to present our offer to the customer. I was wondering if I could let GitHub Copilot interact directly with these documents to help me create and finetune our proposals. Thanks to the rise of the Model Context Protocol (MCP) and a growing list of MCP servers, we can easily do this. But before I dive in the details, let me briefly explain what MCP is all about. What is MCP? MCP, or Model Context Protocol , is an open protocol designed to standardize how applications provide context to large language models (LLMs).  In the documentation they describe it like a USB-C port for AI applications —just as USB-C allows devices to connect seamlessly, MCP enables AI models to integrate...

Microsoft is celebrating its 50 year anniversary and there are giving away a lot of nice presents. One of them is the General Availability of GitHub Copilot Agent mode. I don’t want to repeat everything I said from my preview post so let’s first look at what has changed. Remark : As far as I’m aware Agent Mode is currently only available inside VSCode and not (yet) in Visual Studio First (small) change is that the UI has changed and that the 3 modes are now available through the same dropdown: The basic functionality of Agent Mode in Github Copilot remained the same. So nothing extra to mention there. Remark: If you are in doubt when to use what mode in GitHub Copilot check out this post . Use your own models One new feature I noticed that is not directly related to Agent Mode but was available in the same release is the possibility to use your own models. This can be a hosted model from one of the AI providers or a local model through Ollama. For example, if we wan...