Language Server MCP

A Model Context Protocol (MCP) server that provides rust-analyzer functionality through MCP tools.

Setup

1. Install rust-analyzer:

rustup component add rust-analyzer

2. Build the MCP server:

cargo build --release

Usage with Claude Desktop

Configuration

Claude Desktop uses a configuration file to register MCP servers. The location varies by platform:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add this configuration to your Claude Desktop config file:

{
  "mcpServers": {
    "rust-analyzer": {
      "command": "/path/to/language-server-mcp/target/release/language-server-mcp",
      "args": ["/path/to/your/rust/workspace"]
    }
  }
}

Important Notes:

• Replace /path/to/language-server-mcp with the actual path to this repository
• Replace /path/to/your/rust/workspace with the path to your Rust project
• Ensure rust-analyzer is installed and available in your PATH
• After editing the config, completely quit and restart Claude Desktop
• Look for the MCP server indicator (hammer/tools icon) in the bottom-right corner

Using Desktop Extensions (2025)

For easier installation, you can create a Desktop Extension (.dxt file) that bundles this MCP server. This eliminates the need for manual configuration files.

Usage with Claude Code CLI

Installation

First, ensure Claude Code CLI is installed:

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

Configuration Method 1: CLI Wizard

Use the built-in MCP configuration wizard:

claude mcp add rust-analyzer -s user -- /path/to/language-server-mcp/target/release/language-server-mcp /path/to/your/rust/workspace

Configuration Method 2: Direct Config File

For more control, directly edit the Claude Code configuration file. Create or edit the config file at the appropriate location for your platform and add:

{
  "mcpServers": {
    "rust-analyzer": {
      "type": "stdio",
      "command": "/path/to/language-server-mcp/target/release/language-server-mcp",
      "args": ["/path/to/your/rust/workspace"]
    }
  }
}

Verification

Check that the MCP server is connected:

# Inside Claude Code, run:
/mcp

You should see:

⎿ MCP Server Status ⎿
⎿ • rust-analyzer: connected ⎿

Usage Examples

Once configured, you can use natural language commands with Claude Code:

# Get diagnostics
"Check for errors in src/main.rs"

# Get type information  
"What's the type of the variable at line 25, column 10?"

# Get completions
"What methods are available on this struct?"

# Find definitions
"Where is this function defined?"

# Search workspace
"Find all functions named 'parse' in the codebase"

Available Tools

hover

Get type information and documentation at a specific position.

Parameters:

• file_path: Path to the Rust file
• line: Line number (0-indexed)
• column: Column number (0-indexed)

Example:

{
  "file_path": "/home/user/project/src/main.rs",
  "line": 10,
  "column": 15
}

completion

Get code completions at a specific position.

Parameters:

• file_path: Path to the Rust file
• line: Line number (0-indexed)
• column: Column number (0-indexed)

diagnostics

Get compile errors and warnings for a file.

Parameters:

• file_path: Path to the Rust file

goto_definition

Find the definition location of a symbol.

Parameters:

• file_path: Path to the Rust file
• line: Line number (0-indexed)
• column: Column number (0-indexed)

find_references

Find all references to a symbol.

Parameters:

• file_path: Path to the Rust file
• line: Line number (0-indexed)
• column: Column number (0-indexed)
• include_declaration: Include the declaration in results (optional, default: true)

format_document

Format a Rust file using rustfmt.

Parameters:

• file_path: Path to the Rust file

rename

Rename symbols across the entire workspace safely.

Parameters:

• file_path: Path to the Rust file
• line: Line number (0-indexed)
• column: Column number (0-indexed)
• new_name: The new name for the symbol

code_actions

Get available quick fixes and refactorings at a specific position.

Parameters:

• file_path: Path to the Rust file
• line: Line number (0-indexed)
• column: Column number (0-indexed)

workspace_symbols

Search for symbols across the entire workspace.

Parameters:

• query: Search query string (symbol name pattern)

Example:

{
  "query": "Result"
}

inlay_hints

Get type and parameter hints for a file.

Parameters:

• file_path: Path to the Rust file

expand_macro

Expand Rust macros to see the generated code.

Parameters:

• file_path: Path to the Rust file
• line: Line number (0-indexed)
• column: Column number (0-indexed)

Example Workflow

When using with an AI assistant like Claude:

1. Get diagnostics for a file: "Can you check for errors in src/main.rs?"

2. Get type information: "What's the type of the variable at line 25, column 10 in src/lib.rs?"

3. Get completions: "What methods are available on the object at line 30, column 15?"

4. Search for symbols across the workspace: "Find all functions named 'parse' in the codebase"

5. Get inlay hints for better code understanding: "Show me type hints for src/main.rs"

6. Expand macros to understand generated code: "What does this derive macro expand to at line 15?"

7. Get quick fixes and refactoring suggestions: "What code actions are available for this error?"

8. Safely rename symbols across the workspace: "Rename this function to 'process_data' everywhere it's used"

Testing the Server

You can test the server manually by sending JSON-RPC requests:

# Start the server
cargo run

# In another terminal, send a request
echo '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}' | cargo run

Development

The server consists of three main components:

1. MCP Server (src/main.rs): Handles MCP protocol and tool routing
2. LSP Client (src/lsp_client.rs): Manages rust-analyzer subprocess
3. Tools (src/tools.rs): Defines available MCP tools

To add new LSP features, extend the tool definitions and implement the corresponding handler in call_tool().