phil-scott-78/dotnet-lens-mcp
3.2
If you are the rightful owner of dotnet-lens-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.
A Model Context Protocol (MCP) server that enhances AI assistants with Roslyn-powered code intelligence for C#.
dotnet-lens-mcp
A Model Context Protocol (MCP) server that brings Roslyn-powered code intelligence to your AI assistant.
What does it do?
This MCP server lets AI assistants understand C# code deeply by providing semantic analysis tools:
- Type resolution - Hover over any variable or expression to see its actual type
- Find definitions - Jump to where symbols are defined, just like F12 in VS
- Find references - See everywhere a symbol is used across your solution
- Explore APIs - Get IntelliSense-style member listings at any code position
- Navigate inheritance - Find implementations, derived types, and base classes
- Check build errors - Get real compiler diagnostics for your code
Getting Started
From NuGet (Recommended)
Install with Claude Code
claude mcp add dotnet-lens dnx -- "dotnet-lens-mcp@0.1.0-beta" "--yes"
Manual Configuration
Configure your AI assistant to use the published package:
{
"servers": {
"dotnet-lens": {
"type": "stdio",
"command": "dnx",
"args": ["dotnet-lens-mcp", "--version", "0.1.0-beta", "--yes"]
}
}
}
From Source
For development or testing:
{
"servers": {
"dotnet-lens": {
"type": "stdio",
"command": "dotnet",
"args": ["run", "--project", "/path/to/DotnetLensMcp"]
}
}
}
Usage
Once configured, your AI assistant gains these capabilities:
- First, initialize the workspace - The AI will scan for .sln or .csproj files
- Then use any analysis tools - Get type info, find references, check for errors, etc.
Example prompts:
- "Find all places where this method is called"
- "Show me what methods I can call on this object"
- "What classes implement this interface?"
Requirements
- .NET 9.0 or later
- A C# project or solution to analyze
Feedback & Issues
Found a bug? Not surprised! Please open an issue on our GitHub repository.