Home
Orchard Core
Enhancing OrchardCore AI with the Model Context Protocol (MCP)

Enhancing OrchardCore AI with the Model Context Protocol (MCP)

Artificial Intelligence has come a long way — from simple autocomplete suggestions to powerful, conversational agents. But here’s the catch: even the smartest AI model can’t read your mind.

Context is everything.

Give an LLM a vague prompt, and you’ll get a vague response. But give it rich, structured, real-time context — and it becomes capable of surprising intelligence.

That’s the promise behind the Model Context Protocol (MCP) — and now, thanks to the CrestApps MCP Module, this power is available directly inside your OrchardCore applications.

Whether you're building a chatbot, a content assistant, or a smart dashboard, this new module makes it easy to extend AI behavior by plugging in external tools, services, and even your own backend logic — all without reinventing the wheel.

Let’s explore how MCP is changing the game for AI integrations and how you can get started today.

🤖 What is Model Context Protocol (MCP) and Why It’s a Game Changer

At its core, MCP is a simple but powerful idea: what if there was a universal way for AI models to talk to external systems?

Think of it like the USB-C for AI. With USB-C, you can connect your laptop to power, displays, or storage — no matter the brand. MCP works the same way for AI: it defines a standard protocol that lets large language models connect to tools, data, and services, no matter where they live.

So, what does MCP actually enable?

🧩 Pluggable Tools: Attach prebuilt tools like mcp/time, mcp/calc, or mcp/shell that run locally or in containers.
🌐 Real-Time Context: Provide your AI with live data from internal APIs, databases, or system commands.
🧠 Cross-Model Compatibility: Use the same tools across OpenAI, Ollama, Claude, and more.
🔒 Privacy & Control: Run tools locally or behind your firewall — no need to send data to the cloud.

It’s not just about better responses — it’s about making AI feel integrated into your app’s ecosystem, aware of your user, your data, and your environment.

Learn more: modelcontextprotocol.io/introduction

🧩 Extending OrchardCore AI with MCP

With the CrestApps AI Chat module, you already have a great foundation for building AI-powered experiences in OrchardCore. You can create profiles, hook into your preferred LLM provider, and start chatting with your model right away.

But let’s imagine a common scenario.

🧪 Real-World Example: Why Static AI Isn’t Enough

Let’s say you’ve created an AI profile that connects to your favorite LLM provider — great! Now you type something like:

“What is the current time in Bethlehem, Palestine?”

The model, however, responds with:

“I’m not sure — I don’t have access to real-time data.”

That’s because most language models — even the smartest ones — aren’t connected to the real world. They’re like extremely well-read introverts: lots of knowledge, no access to what’s happening right now.

This is where MCP comes to the rescue.

By connecting your OrchardCore AI profile to a remote or local MCP server — such as one that runs the mcp/time tool — your AI gains the ability to fetch real-time timezone information. The server does the lookup, returns the current time for Palestine, and the model weaves it into its answer naturally.

It feels like magic — but it’s just structured context delivered through MCP.

⚙️ Getting Started with the CrestApps MCP Module

The CrestApps.OrchardCore.AI.Mcp module extends the AI Chat capabilities by allowing your OrchardCore site to interact with MCP servers via:

🌐 Remote MCP Servers using Server-Sent Events (SSE)
🖥 Local MCP Tools via Standard Input/Output (Stdio)

To get started:

Install the CrestApps.OrchardCore.AI.Mcp NuGet package in your web project.
In the OrchardCore dashboard, go to Tools → Features
Enable one or both of:
- Model Context Protocol (MCP) Client
- Model Context Protocol (Local MCP) Client

🌐 Connecting to a Remote MCP Server (SSE Transport)

You can integrate any external MCP Server using Server-Sent Events (SSE), a lightweight, real-time communication method.

🛠 Step-by-Step:

Go to Artificial Intelligence → MCP Connections
Click Add Connection
Under SSE Source, click Add
Fill in details like:

Display Text: Remote AI Time Server
Endpoint: https://localhost:1234/
Additional Headers: (optional)

Click Save

🎯 Then go to AI Profiles and create a new profile using this connection.

🧾 Alternative: Recipe-Based Setup for Remote Server

You can also define the connection via a recipe file:

{
  "steps": [
    {
      "name": "McpConnection",
      "connections": [
        {
          "DisplayText": "Remote AI Tools",
          "Properties": {
            "SseMcpConnectionMetadata": {
              "Endpoint": "https://localhost:1234/",
              "AdditionalHeaders": {}
            }
          }
        }
      ]
    }
  ]
}

This is handy for multi-tenant solutions, automation, or CI/CD pipelines.

🖥 Using the Local MCP Client (Stdio Transport)

Want your AI to work offline or integrate with local tools? Use the Local MCP Client, which runs tools via standard input/output — especially useful with Dockerized tools like:

🕒 mcp/time (returns current time with time zone awareness)
➕ mcp/calc (evaluates math)
🖥 mcp/shell (secure command runner)

🧭 Example: Use `mcp/time` via Docker

Step 1: Install Docker Desktop

Download and install Docker Desktop if you haven't already.

Step 2: Pull the Image

docker pull mcp/time

Step 3: Add the Connection in OrchardCore

Go to AI → MCP Connections
Click Add Connection
Choose Stdio Source and click Add
Enter:

Display Text: Global Time Capabilities
Command: docker
Arguments: ["run", "-i", "--rm", "mcp/time"]

Click Save, and you’re ready to create a profile that uses this connection.

🧾 Recipe-Based Setup for Local Tools

Prefer automation? Here's how to configure the same connection via a recipe:

{
  "steps": [
    {
      "name": "McpConnection",
      "connections": [
        {
          "DisplayText": "Global Time Capabilities",
          "Properties": {
            "StdioMcpConnectionMetadata": {
              "Command": "docker",
              "Arguments": [
                "run",
                "-i",
                "--rm",
                "mcp/time"
              ]
            }
          }
        }
      ]
    }
  ]
}

🧠 More Creative Use Cases

Once MCP is connected, you can get creative with your AI tooling. Here are a few ideas:

🧮 Natural Language Calculator with mcp/calc
Evaluate math queries like “What’s 25% of 480?”
📂 File Search Assistant using mcp/shell
Help users locate files on disk using AI-generated search commands.
🕵️ Internal Monitoring Bot
Connect to internal logs, status scripts, or diagnostic tools.

All of these can be exposed to your AI model using MCP’s consistent, secure protocol.

🌎 Where to Find MCP-Compatible Tools

MCP is growing fast. Explore these hubs for tools and servers:

✅ Wrapping Up

The Model Context Protocol is redefining how AI applications get their context — and with the CrestApps MCP Module, that power is now available to OrchardCore developers with minimal setup and maximum flexibility.

Whether you're looking to add intelligence to your CMS, build AI-powered assistants, or just give your model access to real-world data — MCP offers a clean, scalable, and secure path forward.

Want to try it? 👉 Get started with the CrestApps MCP Module

Let me know if you'd like the Markdown version for publishing or want to add a diagram or code demo!