Windows CLI MCP Server (Enhanced)

MCP server for secure command-line interactions on Windows systems, enabling controlled access to PowerShell, CMD, Git Bash, and Bash shells. It allows MCP clients (like Claude Desktop) to perform operations on your system, similar to Open Interpreter.

This enhanced version includes advanced configuration management, improved security features, and comprehensive testing capabilities.

• Windows CLI MCP Server (Enhanced)

  • Features

  • Usage with Claude Desktop

    • Configuration Setup

  • Configuration

    • Configuration Structure

    • Configuration Locations

    • Default Configuration

    • Configuration Settings

      • Global Settings

        • Security Settings

        • Restriction Settings

        • Path Settings

      • Shell Configuration

        • Basic Shell Configuration

        • Shell-Specific Overrides

        • WSL Configuration

    • Configuration Inheritance

  • API

    • Tools

    • Resources

  • Security Considerations

    • Built-in Security Features

    • Configurable Security Features (Active by Default)

    • Best Practices

  • Using the MCP Inspector for Testing

  • Development and Testing

    • Running Tests

    • Cross-Platform Testing

  • Acknowledgments

  • Development Environment using Dev Containers

    • Prerequisites

    • Getting Started

    • Running Tests in the Dev Container

  • License

Features

• Multi-Shell Support: Execute commands in PowerShell, Command Prompt (CMD), Git Bash, Bash, and WSL

• Modular Architecture: Build only the shells you need for smaller bundle sizes (30-65% reduction)

• Inheritance-Based Configuration: Global defaults with shell-specific overrides

• Shell-Specific Validation: Each shell can have its own security settings and path formats

• Flexible Path Management: Different shells support different path formats (Windows/Unix/Mixed)

• Resource Exposure: View configuration and security settings as MCP resources

• Explicit Working Directory State: The server maintains an active working directory used when execute_command omits workingDir. If the launch directory isn't allowed, this state starts unset and must be set via set_current_directory.

• Optional Initial Directory: Configure initialDir to start the server in a specific directory.

• Security Controls:

  • Command blocking (full paths, case variations)

  • Working directory validation

  • Maximum command length limits

  • Smart argument validation

  • Shell-specific timeout settings

• Configurable:

  • Inheritance-based configuration system

  • Shell-specific security overrides

  • Dynamic tool descriptions based on enabled shells

See the API section for more details on the tools and resources the server provides to MCP clients.

Note: The server will only allow operations within configured directories, with allowed commands.

Related MCP server: Terminal MCP Server

VS Code Extension

A companion VS Code extension in vscode-extension/ simplifies configuring this server. It exposes every CLI option as ordinary VS Code settings (scoped per User and per Workspace) and registers the MCP server with VS Code automatically via the MCP Server Definition Provider API — no hand-edited mcp.json required. It can also generate a config.json or a .vscode/mcp.json on demand. See vscode-extension/README.md.

Modular Shell Architecture

WCLI0 now supports a modular architecture that allows you to build specialized versions containing only the shells you need. This results in significantly smaller bundle sizes and faster startup times.

Build Options

Choose from several pre-configured builds:

# Full build (all shells) - default
npm run build

# Windows-only shells (PowerShell, CMD, Git Bash)
npm run build:windows

# Git Bash only (smallest Windows build)
npm run build:gitbash

# CMD only
npm run build:cmd

# Unix/Linux only (Bash)
npm run build:unix

# Custom combination
INCLUDED_SHELLS=gitbash,powershell npm run build:custom

Bundle Size Comparison

Documentation

For detailed information about the modular architecture:

• Architecture Overview - System design and module structure

• User Guide - How to build and use specialized versions

• API Documentation - Complete API reference for shell plugins

• Migration Guide - Upgrading from previous versions

• Testing Guide - Testing strategies for modular shells

Quick Start with Specialized Builds

If you only need Git Bash:

# Build
npm run build:gitbash

# Use in Claude Desktop config
{
  "mcpServers": {
    "windows-cli": {
      "command": "node",
      "args": ["/path/to/wcli0/dist/index.gitbash-only.js"]
    }
  }
}

macOS and Unix/Linux Support

While wcli0 is primarily designed for Windows, it also supports Unix-based systems (macOS, Linux) with Bash shell integration.

Building for Unix Systems

To build wcli0 for Unix-based systems (macOS, Linux):

# Unix-only build (Bash shell)
npm run build:unix

# The output will be: dist/index.unix-only.js

Starting the Server on macOS

Start the server using npx:

# Start with default settings
npx wcli0 --shell bash

# Start with a configuration file
npx wcli0 --config ./config.mac.json

# Start with specific allowed directories
npx wcli0 --shell bash \
  --allowedDir "/Users/$(whoami)" \
  --allowedDir "/tmp"

macOS Configuration Example

Here's a sample configuration for macOS:

{
  "global": {
    "security": {
      "commandTimeout": 30,
      "enableInjectionProtection": true,
      "restrictWorkingDirectory": true
    },
    "restrictions": {
      "blockedCommands": ["rm -rf /", "dd", "mkfs"],
      "blockedArguments": ["--force", "-rf"],
      "blockedOperators": ["&&", "||", ";", "|"]
    },
    "paths": {
      "allowedPaths": ["/Users/$(whoami)", "/tmp"],
      "initialDir": "/Users/$(whoami)"
    }
  },
  "shells": {
    "bash_auto": {
      "type": "bash_auto",
      "enabled": true
    }
  }
}

Using with Claude Desktop on macOS

Configure Claude Desktop to use wcli0 on macOS:

{
  "mcpServers": {
    "macos-cli": {
      "command": "npx",
      "args": [
        "-y",
        "wcli0",
        "--config",
        "/path/to/config.mac.json"
      ]
    }
  }
}

Important Notes for Unix Systems

• Path Formats: Unix systems use forward slashes (/) and do not support Windows drive letters

• Shell Type: Use bash or bash_auto shell types on Unix systems

• Home Directory: Use $(whoami) or your actual username in paths

• Security Commands: Some blocked commands in the default configuration are Windows-specific (e.g., regedit, format)

CLI Options for macOS

When running on Unix systems, use these CLI options:

Log Management

wcli0 automatically stores command execution logs and provides MCP resources for querying historical output with advanced filtering capabilities.

Output Truncation

By default, command responses show only the last 20 lines to prevent overwhelming long outputs. Full output is always stored and accessible via:

• File-based storage: When logDirectory is configured, logs are saved to files for persistent storage

• In-memory storage: Default behavior using MCP log resources (e.g., cli://logs/commands/{id})

• The get_command_output tool (fallback for hosts that cannot read resources)

Configure truncation settings:

{
  "global": {
    "logging": {
      "maxOutputLines": 20,
      "enableTruncation": true
    }
  }
}

File-Based Log Storage

For persistent logging, configure a log directory:

{
  "global": {
    "logging": {
      "logDirectory": "./logs",
      "exposeFullPath": false
    }
  }
}

Or via CLI:

npx wcli0 --shell gitbash --logDirectory ./logs

When file-based logging is enabled:

• Truncation messages show the file path directly (simpler output)

• Logs persist across server restarts

• No in-memory storage limits apply

• Starting the server with --debug automatically enables file-based logging to your OS temp directory (<temp>/wcli0-debug-logs) when no logDirectory is set, so every command and its output are persisted during debugging sessions.

Security Note: Log files may contain sensitive command output. Ensure the log directory has appropriate permissions.

Log Resources

Access stored command output via MCP resources (in-memory mode):

• cli://logs/list - List all stored command execution logs

• cli://logs/recent?n=10 - Get the N most recent logs

• cli://logs/commands/{id} - Access full output from a specific command

• cli://logs/commands/{id}/range?start=1&end=100 - Query specific line ranges

• cli://logs/commands/{id}/search?q=error&context=3 - Search logs with context

See API Documentation for detailed resource specifications and query parameters.

Example Configuration

{
  "global": {
    "logging": {
      "maxOutputLines": 20,
      "enableTruncation": true,
      "maxStoredLogs": 50,
      "maxLogSize": 1048576,
      "enableLogResources": true,
      "logRetentionMinutes": 1440,
      "logDirectory": "./logs"
    }
  }
}

Usage with Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "windows-cli": {
      "command": "npx",
      "args": ["-y", "wcli0"]
    }
  }
}

For use with a specific config file, add the --config flag:

{
  "mcpServers": {
    "windows-cli": {
      "command": "npx",
      "args": [
        "-y",
        "wcli0",
        "--config",
        "path/to/your/config.json"
      ]
    }
  }
}

Configuration Setup

To get started with configuration:

1. Use a sample configuration:

   • Copy config.examples/config.sample.json for basic setup

   • Copy config.examples/config.development.json for development environments

   • Copy config.examples/config.secure.json for high-security environments

   • Copy config.examples/emptyRestrictions.json to remove all default restrictions

2. Create your own configuration:

   # Copy and customize a sample
   cp config.examples/config.sample.json my-config.json
   
   # Or generate a default config
   npx wcli0 --init-config ./my-config.json

   The server also accepts an --initialDir flag to override the initial working directory defined in your configuration file:

   npx wcli0 --config ./my-config.json --initialDir /path/to/start

   You can override global command limits directly from the CLI:

   npx wcli0 --config ./my-config.json \
     --maxCommandLength 5000 --commandTimeout 60

   You can configure output truncation and logging via CLI:

   npx wcli0 --shell gitbash \
     --maxOutputLines 50 \
     --enableTruncation \
     --enableLogResources \
     --maxReturnLines 1000 \
     --logDirectory ./logs

   Option	Type	Default	Description
   --maxOutputLines	number	20	Maximum output lines before truncation
   --enableTruncation	boolean	true	Enable output truncation
   --enableLogResources	boolean	true	Enable log resources for get_command_output
   --maxReturnLines	number	500	Maximum lines returned by get_command_output
   --logDirectory	string	-	Directory for file-based log storage (instead of in-memory)

   When --logDirectory is configured, command output logs are saved to files instead of in-memory storage. Truncation messages will show the file path for easy access to full output.

   Security Note: Log files may contain sensitive data from command output. Ensure the log directory has appropriate permissions and consider implementing log rotation.

   You can override blocked restrictions directly from the CLI. Pass the option with an empty string to clear defaults:

   npx wcli0 --blockedCommand "" --blockedArgument "" --blockedOperator ""

   Provide the flag multiple times to specify values:

   npx wcli0 --blockedCommand rm --blockedCommand del

   You can also start the server with a specific shell and allowed directories without a configuration file:

   npx wcli0 --shell powershell \
     --allowedDir C:\safe --allowedDir D:\projects

   For WSL shells, you can specify a custom mount location:

npx wcli0 --shell wsl \
  --wslMountPoint /windows/

To disable directory restrictions entirely when no allowed paths are configured, start the server with:

npx wcli0 --allowAllDirs

When started this way, restrictWorkingDirectory is forced on and enableInjectionProtection is disabled to ensure the allowed paths apply without shell injection checks.

If you need to disable safety checks that block command execution for experimentation, you can start the server in unsafe or YOLO modes (not recommended for production):

# YOLO disables all safety checks except allowed working directories
npx wcli0 --yolo

# Fully unsafe removes all safety checks, including directory limits
npx wcli0 --unsafe

Both modes clear blocked commands/arguments/operators and turn off injection protection. YOLO mode leaves working directory restrictions active, while fully unsafe mode disables those restrictions as well. These two flags are mutually exclusive; using both at once will fail.

You can start the server with an HTTP-based transport instead of the default stdio transport, so remote and web-based MCP clients can connect over HTTP. Two HTTP transports are available:

• http -- the modern Streamable HTTP transport (MCP protocol revision 2025-03-26), serving a single /mcp endpoint. This is what current MCP clients default to and is the recommended HTTP transport.

• sse -- the legacy HTTP+SSE transport (MCP protocol revision 2024-11-05), using two endpoints (GET /sse, POST /messages). It is deprecated by the MCP spec in favor of Streamable HTTP and is kept only for compatibility with older clients.

  The modes are mutually exclusive (selected by --transport) and use separate bind settings (--http-* for http, --sse-* for sse).

  # Streamable HTTP on the default host/port (127.0.0.1:9444), serving /mcp
  npx wcli0 --transport http
  
  # Custom port, still bound to localhost
  npx wcli0 --transport http --http-host 127.0.0.1 --http-port 3000
  
  # Legacy HTTP+SSE transport
  npx wcli0 --transport sse --sse-host 127.0.0.1 --sse-port 3000

  Option	Type	Default	Description
  --transport	string	stdio	Transport protocol: stdio, http (Streamable HTTP), or sse (legacy HTTP+SSE)
  --http-host	string	127.0.0.1	Host address for the Streamable HTTP transport (http mode)
  --http-port	number	9444	Port for the Streamable HTTP transport (http mode)
  --http-allowed-origins	string	(none)	Comma-separated browser origins allowed for http mode, in addition to loopback hosts and the bind host (e.g. https://app.example.com,192.168.1.10). Only the host component is compared. Required for browser clients on a wildcard (0.0.0.0) bind.
  --sse-host	string	127.0.0.1	Host address for the legacy SSE transport (sse mode)
  --sse-port	number	9444	Port for the legacy SSE transport (sse mode)
  --sse-allowed-origins	string	(none)	Comma-separated browser origins allowed for sse mode, in addition to loopback hosts and the bind host. Only the host component is compared. Required for browser clients on a wildcard (0.0.0.0) bind.

  When http mode is active, clients use a single /mcp endpoint:

• POST /mcp carries client-to-server JSON-RPC messages. An initialize request with no session id starts a new session; the server returns the assigned id in the Mcp-Session-Id response header, and the client must send that header on every subsequent request.

• GET /mcp opens the optional server-to-client SSE stream for an existing session.

• DELETE /mcp terminates an existing session.

  Sessions are stateful and isolated: each session has its own active working directory, so one client's set_current_directory cannot affect another. Requests carrying an unknown or terminated Mcp-Session-Id are rejected with 404 Not Found. The server logs the bind address and port on startup (with --debug).

  When sse mode is active, clients instead connect via GET /sse to open an SSE stream and send messages via POST /messages?sessionId=<id>.

  Configuring Streamable HTTP entirely with CLI parameters (no config file). Every transport and operational setting can be supplied as an input parameter, so the server can run as a Streamable HTTP server without any config file:

  npx wcli0 \
    --transport http \
    --http-host 127.0.0.1 \
    --http-port 9444 \
    --http-allowed-origins "https://app.example.com,192.168.1.10" \
    --shell gitbash \
    --allowedDir "D:/work/project" \
    --commandTimeout 60 \
    --debug

  CLI parameters also take precedence over a config file, so the same --http-* flags override the corresponding transport fields when a --config file is also passed (see the transport config section).

  Both HTTP transports validate the request Origin header to mitigate DNS-rebinding attacks: requests whose Origin is not a loopback host, the configured bind host, or one of the configured allowed origins (--http-allowed-origins / --sse-allowed-origins) are rejected with 403 Forbidden, while non-browser clients that send no Origin are allowed. Allowed browser origins receive CORS headers, and OPTIONS preflight requests are answered with 204.

  Security: Neither HTTP transport has built-in authentication, and both expose command-execution tools. Keep the server bound to 127.0.0.1 (the default) for local use. Binding to 0.0.0.0 or any non-loopback address exposes those tools to every host that can reach the port; only do so behind an authenticated reverse proxy or equivalent access control. Origin validation alone does not authenticate non-browser clients.

  Wildcard binds and browser origins: When binding to a wildcard address (0.0.0.0 / ::), the bind host is not a usable origin to compare against, so browser clients reaching the server through its real LAN address (or a reverse proxy whose public hostname differs from the bind host) are rejected unless their origin is listed in --http-allowed-origins / --sse-allowed-origins (or the transport.httpAllowedOrigins / transport.sseAllowedOrigins config arrays). Non-browser clients are unaffected, since they send no Origin.

1. Update your Claude Desktop configuration to use your config file:

   {
     "mcpServers": {
       "windows-cli": {
         "command": "npx",
         "args": [
           "-y",
           "wcli0",
           "--config",
           "./my-config.json"
         ]
       }
     }
   }

After configuring, you can:

• Execute commands directly using the available tools

• View server configuration and security settings in the Resources section

• Access shell-specific configurations and capabilities

Configuration

The server uses an inheritance-based configuration system where global defaults can be overridden by shell-specific settings.

Configuration Structure

{
  "global": {
    "security": {
      "maxCommandLength": 2000,
      "commandTimeout": 30,
      "enableInjectionProtection": true,
      "restrictWorkingDirectory": true
    },
    "restrictions": {
      "blockedCommands": ["format", "shutdown"],
      "blockedArguments": ["--exec", "-e"],
      "blockedOperators": ["&", "|", ";", "`"]
    },
    "paths": {
      "allowedPaths": ["/home/user", "/tmp"],
      "initialDir": "/home/user"
    }
  },
  "shells": {
    "powershell": {
      "type": "powershell",
      "enabled": true,
      "executable": {
        "command": "powershell.exe",
        "args": ["-NoProfile", "-NonInteractive", "-Command"]
      },
      "overrides": {
        "security": {
          "commandTimeout": 45
        },
        "restrictions": {
          "blockedCommands": ["Remove-Item", "Format-Volume"]
        }
      }
    },
    "wsl": {
      "type": "wsl",
      "enabled": true,
      "executable": {
        "command": "wsl.exe",
        "args": ["-e"]
      },
      "wslConfig": {
        "mountPoint": "/mnt/",
        "inheritGlobalPaths": true
      }
    }
  }
}

Configuration Locations

The server looks for configuration files in the following order:

1. Path specified via --config command line argument

2. win-cli-mcp.config.json in the current working directory

3. ~/.win-cli-mcp/config.json in user's home directory

If no configuration file is found, the server will use a default (restricted) configuration.

Default Configuration

Note: The default configuration is designed to be restrictive and secure. Find more details on each setting in the Configuration Settings section.

For a complete reference of all default values, see docs/defaults.md.

{
  "global": {
    "security": {
      "maxCommandLength": 2000,
      "commandTimeout": 30,
      "enableInjectionProtection": true,
      "restrictWorkingDirectory": true
    },
    "restrictions": {
      "blockedCommands": [
        "rm", "del", "rmdir", "format", "shutdown", "restart",
        "reg", "regedit", "net", "netsh", "takeown", "icacls"
      ],
      "blockedArguments": [
        "--exec", "-e", "/c", "-enc", "-encodedcommand",
        "-command", "--interactive", "-i", "--login", "--system"
      ],
      "blockedOperators": ["&", "|", ";", "`"]
    },
    "paths": {
      "initialDir": null
    }
  },
  "shells": {
    "powershell": {
      "type": "powershell",
      "enabled": true,
      "executable": {
        "command": "powershell.exe",
        "args": ["-NoProfile", "-NonInteractive", "-Command"]
      }
    },
    "cmd": {
      "type": "cmd",
      "enabled": true,
      "executable": {
        "command": "cmd.exe",
        "args": ["/c"]
      }
    },
    "gitbash": {
      "type": "gitbash",
      "enabled": true,
      "executable": {
        "command": "C:\\Program Files\\Git\\bin\\bash.exe",
        "args": ["-c"]
      }
    }
  }
}

Configuration Settings

The configuration file uses an inheritance system with two main sections: global and shells.

Global Settings

Global settings provide defaults that apply to all shells unless overridden.

Security Settings

{
  "global": {
    "security": {
      // Maximum allowed length for any command
      "maxCommandLength": 2000,

      // Command execution timeout in seconds
      "commandTimeout": 30,

      // Enable protection against command injection
      "enableInjectionProtection": true,

      // Restrict commands to allowed working directories
      "restrictWorkingDirectory": true
    }
  }
}

Restriction Settings

{
  "global": {
    "restrictions": {
      // Commands to block - blocks both direct use and full paths
      "blockedCommands": ["rm", "format", "shutdown"],

      // Arguments to block across all commands
      "blockedArguments": ["--exec", "-e", "/c"],

      // Operators to block in commands
      "blockedOperators": ["&", "|", ";", "`"]
    }
  }
}

Path Settings

{
  "global": {
    "paths": {
      // Directories where commands can be executed
      "allowedPaths": ["/home/user", "/tmp", "C:\\Users\\username"],

      // Initial working directory (null = use launch directory)
      "initialDir": "/home/user",

      // Whether to restrict working directories
      "restrictWorkingDirectory": true
    }
  }
}

If the allowedPaths array is omitted from your configuration file, no default directories are automatically allowed. When restrictWorkingDirectory is enabled, only the initialDir (if specified) will be added to the allowed paths list. Use the --allowAllDirs flag when launching the server to automatically disable restrictWorkingDirectory if no allowed paths or initialDir are set.

Shell Configuration

Each shell can be individually configured and can override global settings. Each shell entry must include a type field indicating the shell. Valid values are powershell, cmd, gitbash, bash, and wsl.

Basic Shell Configuration

{
  "shells": {
    "powershell": {
      "type": "powershell",
      "enabled": true,
      "executable": {
        "command": "powershell.exe",
        "args": ["-NoProfile", "-NonInteractive", "-Command"]
      }
    }
  }
}

Shell-Specific Overrides

{
  "shells": {
    "powershell": {
      "type": "powershell",
      "enabled": true,
      "executable": {
        "command": "powershell.exe",
        "args": ["-NoProfile", "-NonInteractive", "-Command"]
      },
      "overrides": {
        "security": {
          "commandTimeout": 45,
          "maxCommandLength": 3000
        },
        "restrictions": {
          "blockedCommands": ["Remove-Item", "Format-Volume"],
          "blockedOperators": ["|", "&"]
        }
      }
    }
  }
}

WSL Configuration

WSL shells have additional configuration options for path mapping:

{
  "shells": {
    "wsl": {
      "type": "wsl",
      "enabled": true,
      "executable": {
        "command": "wsl.exe",
        "args": ["-e"]
      },
      "wslConfig": {
        "mountPoint": "/mnt/",
        "inheritGlobalPaths": true
      }
    }
  }
}

You can override the mount point at startup using the --wslMountPoint CLI flag.

Configuration Inheritance

The transport section can also be set in the config file. For the Streamable HTTP transport (http mode):

{
  "transport": {
    "mode": "http",
    "httpHost": "127.0.0.1",
    "httpPort": 9444,
    "httpAllowedOrigins": ["https://app.example.com", "192.168.1.10"]
  }
}

For the legacy HTTP+SSE transport (sse mode):

{
  "transport": {
    "mode": "sse",
    "sseHost": "127.0.0.1",
    "ssePort": 9444,
    "sseAllowedOrigins": ["https://app.example.com", "192.168.1.10"]
  }
}

The *AllowedOrigins lists are optional and default to an empty list. Each entry is an origin URL or a bare host; only the host component is compared (case-insensitively).

CLI flags override config-file values: --transport, --http-host, --http-port, --http-allowed-origins, --sse-host, --sse-port, and --sse-allowed-origins.

The inheritance system works as follows:

1. Global defaults are applied to all shells

2. Shell-specific overrides replace or extend global settings

3. Array settings (like blockedCommands) override defaults when provided. Specifying an empty array removes all default entries for that setting.

4. Object settings are deep-merged

5. Primitive settings are replaced

Example of inheritance in action:

{
  "global": {
    "security": { "commandTimeout": 30 },
    "restrictions": { "blockedCommands": ["rm", "format"] }
  },
  "shells": {
    "powershell": {
      "type": "powershell",
      "overrides": {
        "security": { "commandTimeout": 45 },
        "restrictions": { "blockedCommands": ["Remove-Item"] }
      }
    }
  }
}

Results in PowerShell having:

• commandTimeout: 45 (overridden)

• blockedCommands: ["Remove-Item"] (overrides defaults)

To completely remove defaults for a given restriction, provide an empty array:

{
  "global": {
    "restrictions": {
      "blockedCommands": [],
      "blockedArguments": [],
      "blockedOperators": []
    }
  },
  "shells": {
    "powershell": {
      "type": "powershell",
      "overrides": {
        "restrictions": { "blockedCommands": [] }
      }
    }
  }
}

Environment Profiles

Named environment profiles let a single server instance run the same CLI tool under different environment variable sets, selected per call via the optional profile parameter on execute_command. A common use case is testing the same SQL against different sqlplus versions, where each version needs its own ORACLE_HOME, TNS_ADMIN, and a PATH that points at that version's bin.

Profiles are defined under an optional top-level profiles map. Each entry accepts:

{
  "profiles": {
    "ora19": {
      "description": "Oracle 19c sqlplus client",
      "allowedShells": ["cmd", "powershell"],
      "env": {
        "ORACLE_HOME": "C:\\oracle\\product\\19.0.0\\client",
        "TNS_ADMIN": "C:\\oracle\\product\\19.0.0\\client\\network\\admin",
        "PATH": "C:\\oracle\\product\\19.0.0\\client\\bin;${PATH}"
      }
    }
  }
}

Behavior:

• When a profile is selected, its env map is merged over the server's environment ({ ...process.env, ...profileEnv }) before the command runs.

• ${VAR} is replaced with the server environment value of VAR; an undefined reference resolves to an empty string. This is how PATH is prepended ("C:\\oracle\\product\\19.0.0\\client\\bin;${PATH}").

• Profiles are validated at load time: env must be a non-empty string-to-string map and every allowedShells entry must be a known shell. Invalid profiles abort startup with a descriptive error.

• Selecting an unknown profile, or a profile whose allowedShells excludes the requested shell, returns an InvalidParams error.

• When no profiles are configured, behavior is unchanged and the profile parameter is not exposed.

A complete example is provided in config.examples/profiles.json. See Configuration Examples for more.

API

Tools

• execute_command

  • Execute a command in the specified shell

  • Inputs:

    • shell (string): Shell to use ("powershell", "cmd", "gitbash", "bash", or "wsl")

    • command (string): Command to execute

    • workingDir (optional string): Working directory

    • maxOutputLines (optional number): Maximum output lines to return (1-10,000). Overrides global setting.

    • timeout (optional number): Command timeout in seconds (1-3,600). Overrides global setting.

    • profile (optional string): Named environment profile to apply for this command. Must name a configured profile (see Environment Profiles). Only present in the schema when profiles are configured; omit to run with the server's default environment.

  • Returns command output as text, or error message if execution fails

  • If workingDir is omitted, the command runs in the server's active working directory. If this has not been set, the tool returns an error.

• get_current_directory

  • Get the server's active working directory

  • If the directory is not set, returns a message explaining how to set it

• set_current_directory

  • Set the server's active working directory

  • Inputs:

    • path (string): Path to set as current working directory

  • Returns confirmation message with the new directory path, or error message if the change fails

• get_config

  • Get the windows CLI server configuration

  • Returns the server configuration as a JSON string (excluding sensitive data)

• validate_directories

  • Check if specified directories are within allowed paths

  • Only available when restrictWorkingDirectory is enabled in configuration

  • Inputs:

    • directories (array of strings): List of directory paths to validate

  • Returns success message if all directories are valid, or error message detailing which directories are outside allowed paths

Resources

• cli://config

  • Returns the main CLI server configuration (excluding sensitive data like blocked command details if security requires it).

• cli://logs/list

  • List all stored command execution logs with metadata

• cli://logs/recent?n={count}

  • Get the N most recent command logs (default: 5)

• cli://logs/commands/{id}

  • Access full output from a specific command execution

• cli://logs/commands/{id}/range?start={n}&end={m}

  • Query specific line ranges from a log (supports negative indices)

• cli://logs/commands/{id}/search?q={pattern}&context={n}&occurrence={n}

  • Search logs with regex patterns and context lines

Security Considerations

This server allows external tools to execute commands on your system. Exercise extreme caution when configuring and using it.

Built-in Security Features

• Path Restrictions: Commands can only be executed in specified directories (allowedPaths) if restrictWorkingDirectory is true.

• Command Blocking: Defined commands and arguments are blocked to prevent potentially dangerous operations (blockedCommands, blockedArguments).

• Injection Protection: Common shell injection characters (;, &, |, `) are blocked in command strings if enableInjectionProtection is true.

• Timeout: Commands are terminated if they exceed the configured timeout (commandTimeout).

• Input validation: All user inputs are validated before execution

• Shell process management: Processes are properly terminated after execution or timeout

Configurable Security Features (Active by Default)

• Working Directory Restriction (restrictWorkingDirectory): HIGHLY RECOMMENDED. Limits command execution to safe directories.

• Injection Protection (enableInjectionProtection): Recommended to prevent bypassing security rules.

Best Practices

• Minimal Allowed Paths: Only allow execution in necessary directories.

• Restrictive Blocklists: Block any potentially harmful commands or arguments.

• Regularly Review Logs: Check the command history for suspicious activity.

• Keep Software Updated: Ensure Node.js, npm, and the server itself are up-to-date.

Using the MCP Inspector for Testing

Use the Inspector to interactively test this server with a custom config file. Pass any server flags after --:

# Inspect with built server and test config
npx @modelcontextprotocol/inspector -- node dist/index.js --config tests/config.json

# Or test the published package
npx @modelcontextprotocol/inspector wcli0 -- --config tests/config.json

Development and Testing

This project requires Node.js 18 or later.

Running Tests

# Install dependencies
npm install

# Run all tests
npm test

# Run specific test suites
npm run test:validation    # Path validation tests
npm run test:wsl          # WSL emulation tests
npm run test:integration  # Integration tests
npm run test:async        # Async operation tests

# Run tests with coverage
npm run test:coverage

# Debug open handles
npm run test:debug

Cross-Platform Testing

The project uses a Node.js-based WSL emulator (scripts/wsl-emulator.js) to enable testing of WSL functionality on all platforms. This allows the test suite to run successfully on both Windows and Linux environments.

Acknowledgments

This project is based on the excellent work by SimonB97 in the win-cli-mcp-server repository. Due to significant configuration differences and architectural changes that made merging back to the source repository challenging, this has been maintained as a separate fork with enhanced features and extensive modifications.

Key enhancements in this version:

• Enhanced inheritance-based configuration system

• Improved WSL support with cross-platform testing

• Advanced security features and path validation

• Comprehensive test coverage with Node.js-based WSL emulation

• Extended documentation and configuration examples

We gratefully acknowledge SimonB97's foundational work that made this project possible.

Development Environment using Dev Containers

This project includes a Dev Container configuration, which allows you to use a Docker container as a fully-featured development environment. This ensures consistency and makes it easy to get started with development and testing.

Prerequisites

• Docker Desktop installed and running.

• Visual Studio Code installed.

• The Dev Containers extension installed in VS Code.

Getting Started

1. Clone this repository to your local machine.

2. Open the repository in Visual Studio Code.

3. When prompted "Reopen in Container", click the button. (If you don't see a prompt, you can open the Command Palette (Ctrl+Shift+P or Cmd+Shift+P) and select "Dev Containers: Reopen in Container".)

4. VS Code will build the dev container image (as defined in .devcontainer/devcontainer.json and Dockerfile) and start the container. This might take a few minutes the first time.

5. Once the container is built and started, your VS Code will be connected to this environment. The postCreateCommand (npm install) will ensure all dependencies are installed.

Running Tests in the Dev Container

After opening the project in the dev container:

1. Open a new terminal in VS Code (it will be a terminal inside the container).

2. Run the tests using the command:

   npm test

This setup mirrors the environment used in GitHub Actions for tests, ensuring consistency between local development and CI.

License

This project is licensed under the MIT License. See the LICENSE file for details.