MCP File Server
About
A Docker-based MCP server that allows AI assistants like Claude for Desktop to read and write files from your local file system through a set of file operation tools.
Explore
- List files and directories
- Read file contents
- Write content to files
- Delete files and directories
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
MCP File ServerCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
- Docker installed on your system
- Git (optional, for cloning the repository)
list_files
List all files in a directory
read_file
Read the contents of a file
write_file
Write content to a file
delete_file
Delete a file or directory
The following tools are available through this MCP server:
- list_files: List all files in a directory
- read_file: Read the contents of a file
- write_file: Write content to a file
- delete_file: Delete a file or directory
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"mcp file server": {
"mcp-file-server-abhishekloiwal": {
"command": "docker",
"args": [
"build",
"-t",
"mcp-file-server",
"."
]
}
}
}
}McpServers
{
"mcp-file-server-abhishekloiwal": {
"command": "docker",
"args": [
"build",
"-t",
"mcp-file-server",
"."
]
}
}An MCP server for reading and writing files from your local file system. This server can be used with Claude for Desktop or any other MCP client to provide file system access for AI assistants.
Features
- List files and directories
- Read file contents
- Write content to files
- Delete files and directories
Prerequisites
- Docker installed on your system
- Git (optional, for cloning the repository)
Setup and Deployment
Option 1: Using Docker Compose (Recommended)
1. Clone this repository:
git clone https://github.com/abhishekloiwal/mcp-file-server.git
cd mcp-file-server
2. Edit the docker-compose.yml file to update the volume mount path if needed. By default, it's set to:
volumes:
- /Users/abhishekloiwal/CascadeProjects/ClaudeProjects:/data
Replace with your desired local path if different.
3. Deploy with Docker Compose:
docker-compose up -d
Option 2: Using Docker directly
1. Clone the repository:
git clone https://github.com/abhishekloiwal/mcp-file-server.git
cd mcp-file-server
2. Build the Docker image:
docker build -t mcp-file-server .
3. Run the container with your local directory mounted:
docker run -d --name mcp-file-server -v /Users/abhishekloiwal/CascadeProjects/ClaudeProjects:/data mcp-file-server
Replace the path with your desired local directory path.
Connecting to Claude for Desktop
1. Create or update your Claude for Desktop configuration file at:
- Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
- Windows: %AppData%\Claude\claude_desktop_config.json
2. Add the mcp-file-server to your configuration:
{
"mcpServers": {
"file-server": {
"command": "docker",
"args": ["exec", "-i", "mcp-file-server", "python", "server.py"]
}
}
}
3. Restart Claude for Desktop.
4. You should now see the file-server tools available in Claude.
Available Tools
The following tools are available through this MCP server:
- list_files: List all files in a directory
- read_file: Read the contents of a file
- write_file: Write content to a file
- delete_file: Delete a file or directory
License
MIT
Troubleshooting
- If Claude for Desktop doesn't connect to the server, check the Docker container status:
docker ps -a | grep mcp-file-server
- View server logs:
docker logs mcp-file-server
- Make sure the volume is correctly mounted:
docker inspect mcp-file-server | grep -A 10 Mounts
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
