Get the FREE Ultimate OpenClaw Setup Guide →

fast-filesystem

A high-performance Model Context Protocol (MCP) server that provides secure filesystem access for Claude and other AI assistants.

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio efforthye-fast-filesystem-mcp npx -y fast-filesystem-mcp

How to use

The fast-filesystem MCP server provides a collection of Claude-optimized filesystem tools that enable high-performance read/write operations, large-file handling, and advanced file management. It exposes utilities for reading and writing files with chunking, listing and manipulating directories, searching by name or content, and performing complex edits with built-in backup and recovery features. To use it, configure Claude Desktop to load the fast-filesystem MCP server as shown in the Quick Start. Once active, you can invoke tools such as fast_read_file, fast_write_file, fast_large_write_file, fast_list_directory, fast_search_files, fast_copy_file, fast_move_file, fast_delete_file, and the range of editing tools (fast_edit_file, fast_edit_multiple_blocks, fast_extract_lines, etc.). These tools are designed to be memory-efficient, support paging and streaming for large datasets, and integrate with the backup and retry features to safeguard data during automated edits.

Typical workflows include reading or modifying large configuration files with chunked reads, performing safe line-based edits with built-in backups, and synchronizing directories while monitoring progress. The available tools span core file operations, complex file management, archive handling, directory synchronization, editing utilities, and system-like utilities (disk usage, allowed directories). For best results, enable backup and debug options as needed to diagnose issues without impacting Claude’s JSON-RPC parsing in responses.

How to install

Prerequisites:\n- Node.js and npm (npx is included with npm).\n- Internet access to install the MCP package from the npm registry.\n\nStep 1: Install the MCP server globally (recommended)\nbash\nnpm install -g fast-filesystem-mcp\n\n\nStep 2: Verify installation and version\nbash\nfast-filesystem-mcp --version\n\n\nStep 3: Configure Claude Desktop to load the MCP server\n- Add to your Claude Desktop config (example under mcpServers):\njson\n{\n "mcpServers": {\n "fast-filesystem": {\n "command": "npx",\n "args": ["-y", "fast-filesystem-mcp"]\n }\n }\n}\n\n\nStep 4: Optional backup and logging configuration\n- To enable backups for edits, set CREATE_BACKUP_FILES to true in the environment:\njson\n{\n "mcpServers": {\n "fast-filesystem": {\n "command": "npx",\n "args": ["-y", "fast-filesystem-mcp"],\n "env": {\n "CREATE_BACKUP_FILES": "true"\n }\n }\n }\n}\n\n\nStep 5: Test a basic operation\n- Try listing a directory or reading a small file to confirm the MCP server responds correctly. Ensure Claude Desktop can reach the MCP endpoint and that the server process remains responsive.

Additional notes

  • Backup behavior: By default, backups are created before modifications if the backup features are enabled. Backups are timestamped (e.g., file.txt.backup.1234567890) to reduce risk of data loss.\n- Debug/logging: You can enable verbose MCP logging by setting DEBUG_MCP=true or MCP_DEBUG=true. Logs can be written to a file via MCP_LOG_FILE. Silent errors can be enabled with MCP_SILENT_ERRORS to minimize error messages in responses.\n- Performance: The MCP emphasizes streaming and chunked operations to handle large files efficiently. Use fast_large_write_file and fast_read_file variants for large data.\n- Compatibility: The tooling is optimized for Claude Desktop; avoid extremely verbose outputs unless debugging to prevent JSON parsing issues.\n- Availability: The tools cover a wide range of file operations, directory synchronization, archive handling, and advanced edits. If you encounter issues, check backup availability, environment variable configurations, and ensure the MCP server version matches the documented interfaces.

Related MCP Servers

Sponsor this space

Reach thousands of developers