# KindlyGuard Wrap Configuration
The wrap command now supports configuration files to customize auto-wrapping behavior for AI CLI tools.
## Configuration File Location
The configuration file is stored at: `~/.kindlyguard/wrap.toml`
## Configuration Options
### Basic Settings
- `enabled` (bool): Enable or disable auto-wrapping globally
- `mode` (string): Protection mode - either "warning" or "blocking"
- `"warning"`: Shows threat warnings but allows execution
- `"blocking"`: Prevents execution when threats are detected
- `server` (string): URL of the KindlyGuard server for threat detection
- `verbose` (bool): Show detailed threat information
- `log_sessions` (bool): Save wrapped session logs to disk
- `log_directory` (string, optional): Directory for session logs
### Command Lists
- `commands`: List of commands to auto-wrap (includes defaults like "claude", "openai", etc.)
- `custom_commands`: Additional user-defined commands to wrap
## Usage Examples
### Initialize Configuration
Create a default configuration file:
```bash
kindly-tools wrap --init
```
### Show Current Configuration
Display the current wrap configuration:
```bash
kindly-tools wrap --config
```
### Add a Custom Command
Add a new command to the wrap list:
```bash
kindly-tools wrap --add my-ai-tool
```
### Remove a Command
Remove a command from the wrap list:
```bash
kindly-tools wrap --remove gemini
```
### Run with Auto-Wrap
When configuration is enabled, just run any configured command normally:
```bash
claude "Help me write a script"
# KindlyGuard will automatically wrap and protect this command
```
### Override Configuration
Force wrap any command regardless of configuration:
```bash
kindly-tools wrap -- any-command --args
```
## Example Configuration
```toml
# Enable auto-wrapping
enabled = true
# Use blocking mode for maximum security
mode = "blocking"
# Local KindlyGuard server
server = "http://localhost:8080"
# Show detailed information
verbose = true
# Log all sessions
log_sessions = true
log_directory = "/home/user/.kindlyguard/logs"
# Default AI commands
commands = [
"claude",
"openai",
"gemini",
"anthropic",
"gpt",
"ai",
"llm"
]
# Company-specific AI tools
custom_commands = [
"company-llm",
"internal-ai-cli"
]
```
## Security Modes
### Warning Mode
- Displays threat warnings to the user
- Allows the command to proceed
- Useful for monitoring and awareness
### Blocking Mode
- Prevents potentially dangerous input from reaching the AI
- Blocks execution when threats are detected
- Recommended for production environments
## Command Detection
The wrap configuration intelligently handles:
- Simple command names (e.g., `claude`)
- Full paths (e.g., `/usr/bin/claude`)
- Relative paths (e.g., `./claude`)
Only the basename is checked against the configuration list.