MCP Filesystem Server (Encoding Supported)

A Model Context Protocol (MCP) server for filesystem operations, optimized for intelligent interaction with large files and covering various text encodings (like UTF-8, EUC-KR).

Quick Setup with uvx

You can easily run this MCP server using uvx (the execution tool provided by uv). To configure this server in your MCP client (such as Claude Desktop or Antigravity), add the following to your configuration file:

MCP Client Configuration

Add the following JSON to your MCP settings file (For Claude Desktop):

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

{
  "mcpServers": {
    "mcp-filesystem-encoding": {
      "command": "uvx",
      "args": [
        "mcp-filesystem-encoding",
        "C:\\Users\\username\\Desktop\\Test",
        "C:\\Users\\username\\Documents\\Projects"
      ]
    }
  }
}

Note: The arguments provided after the package name act as your Allowed Root Directories.

• Multiple Paths Supported: You can specify as many root directories as you need by adding them as separate string arguments in the args array (e.g., "C:\\dir1", "C:\\dir2", etc.).
• Security Boundary: The MCP server will strictly restrict its file operations (read, write, search) to only these listed directories and their subdirectories.

Important Variables & Usage Guidelines

1. Using the path Variable (Config & Tools)

• At Server Setup: Provide the allowed root paths in your args list (as shown above).
• During Tool Execution: When the MCP client (AI) uses tools like read_file or search_files, it must specify the fully qualified absolute path (e.g., C:\Users\username\Desktop\Test\file.txt). If the AI requests a path outside of the configured root directories, the server will deny access.

2. The encoding Option

This extended MCP server natively supports reading, writing, and searching files across different encodings, preventing character corruption in non-UTF-8 local environments (such as Korean euc-kr / cp949, or Japanese shift-jis).

The encoding parameter is optional (defaults to "utf-8") and is readily available across the following tool operations:

• Examine Content: read_file, read_multiple_files, head_file, tail_file
• Modify Content: write_file, edit_file
• Discover Content: search_files (for precise text matching), grep_files (advanced Regex searches requiring decoding of target files)

Usage Example in MCP Call:

// Example: Reading a legacy Korean document
{
  "path": "C:\\Users\\Username\\Documents\\legacy_document.txt",
  "encoding": "euc-kr"
}

Available Tools Summary

• Basic IO: read_file, write_file, edit_file
• Search: search_files, grep_files
• File System: list_directory, create_directory, directory_tree
• Analytics: compare_files, find_duplicate_files

(All internal tools strictly enforce the path policy boundaries and seamlessly support the encoding option to empower your environment without breaking localized text!)