AI Quick Start¶

Orchard Core is a large, modular framework, and the fastest way to become productive with it is to let an AI assistant work alongside you. Instead of searching the documentation by hand, you can ask an assistant to scaffold modules, write recipes, generate content type migrations, and explain how a feature works — all grounded in accurate, up-to-date Orchard Core knowledge.

The key to accurate answers is the official Orchard Core MCP server. It gives AI assistants direct, tool-based access to the Orchard Core documentation and APIs, so responses reflect how Orchard Core actually works rather than guesses.

What is the MCP server?¶

Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools and data sources. The Orchard Core MCP server exposes tools that let an assistant search and read the Orchard Core documentation so it can answer questions and generate code with the correct APIs, module names, and conventions.

The server is hosted at:

https://ai.orchardcore.net

Point any MCP-capable AI client at this URL, and the assistant gains access to the Orchard Core tools automatically.

Connecting your AI assistant¶

The exact steps depend on the client you use, but they all point to the same remote endpoint, https://ai.orchardcore.net. Below are the most common clients. Since Orchard Core is a .NET project, GitHub Copilot is listed first.

GitHub Copilot CLI¶

The GitHub Copilot CLI brings a Copilot coding agent to your terminal. Install it with npm (requires Node.js):

npm install -g @github/copilot

Start it by running copilot, then register the MCP server:

copilot
/mcp add

When prompted, choose an HTTP server, name it orchardcore, and use the URL https://ai.orchardcore.net. Alternatively, add it to your Copilot CLI mcp-config.json:

{
  "mcpServers": {
    "orchardcore": {
      "type": "http",
      "url": "https://ai.orchardcore.net"
    }
  }
}

GitHub Copilot (online)¶

For GitHub Copilot on github.com (Copilot Chat and the Copilot coding agent), add the Orchard Core MCP server in your repository or organization settings under Settings → Copilot → Coding agent → MCP configuration:

{
  "mcpServers": {
    "orchardcore": {
      "type": "http",
      "url": "https://ai.orchardcore.net"
    }
  }
}

Once saved, Copilot can call the Orchard Core tools when working on issues and pull requests.

Visual Studio Code (GitHub Copilot)¶

Create or edit .vscode/mcp.json in your workspace:

{
  "servers": {
    "orchardcore": {
      "type": "http",
      "url": "https://ai.orchardcore.net"
    }
  }
}

Open the Chat view in Agent mode and confirm the orchardcore server is listed among the available tools.

Claude Code¶

Claude Code is a terminal-based AI coding assistant. If running claude returns an error such as 'claude' is not recognized as an internal or external command (or command not found on macOS/Linux), it is not installed yet. Install it with npm (requires Node.js):

npm install -g @anthropic-ai/claude-code

Verify the installation:

claude --version

Then add the Orchard Core MCP server:

claude mcp add --transport http orchardcore https://ai.orchardcore.net

Start a session and ask Orchard Core questions — the assistant will call the server's tools when it needs Orchard Core knowledge.

Claude Desktop¶

Edit your claude_desktop_config.json (Settings → Developer → Edit Config) and add the server under mcpServers:

{
  "mcpServers": {
    "orchardcore": {
      "type": "http",
      "url": "https://ai.orchardcore.net"
    }
  }
}

Restart Claude Desktop for the change to take effect.

Cursor¶

Edit .cursor/mcp.json in your project (or the global ~/.cursor/mcp.json):

{
  "mcpServers": {
    "orchardcore": {
      "url": "https://ai.orchardcore.net"
    }
  }
}

Other clients¶

Any client that supports remote (HTTP) MCP servers can connect. Look for an "MCP", "Model Context Protocol", or "Tools/Connectors" setting and register a new HTTP server with the URL https://ai.orchardcore.net.

Power your assistant with AI Skills¶

Beyond the MCP server, you can teach your assistant repeatable, Orchard Core–specific workflows using AI Skills. The CrestApps.AgentSkills project provides a curated set of skills and plugins for building Orchard Core solutions with AI.

Note

CrestApps.AgentSkills is a community project. It is maintained by the community and is not an official part of Orchard Core.

You can install these as plugins for either the Claude CLI or the GitHub Copilot CLI, which is an alternative way to power your AI tool with reusable skills. Follow the instructions in the CrestApps.AgentSkills repository to install the plugin for your CLI of choice.

Tip

Combine the Orchard Core MCP server with the AI Skills plugins for the best results: the MCP server keeps the assistant grounded in accurate documentation, while the skills guide it through common Orchard Core tasks step by step.

Building on top of Orchard Core with AI¶

Once connected, use natural language to build on top of your Orchard Core workflow. Because the assistant can consult the official documentation through the MCP server, its suggestions stay aligned with the framework's conventions. Some examples of what to ask:

Scaffold a module or theme — "Create a new Orchard Core module that adds a Product content type with a price field."
Write recipes — "Generate a recipe that creates a blog content type, a menu, and a home page."
Content modeling — "Write a data migration that adds an AutoroutePart and a MarkdownBodyPart to my Article type."
Explain features — "How do display drivers and shapes work, and how do I customize the rendering of a part?"
Extend behavior — "Show me how to handle a content event so I can send a notification when an item is published."

Tip

Give the assistant context about your project (the modules you have enabled, the content types you use, and your target Orchard Core version) so its answers and generated code fit your solution.