CLI Options

Global Options

These options are available across all Cline CLI commands.

Mode Options

boolean

Description: Run in Act mode (default)Behavior: Cline actively uses tools to accomplish tasks - reading files, writing code, executing commands, and using a headless browser.Example:

cline --act "Add unit tests to the authentication module"

boolean

Description: Run in Plan modeBehavior: Cline gathers information and creates a detailed plan before implementation. He explores the codebase, asks clarifying questions, and presents a strategy for approval.Example:

cline --plan "Design a REST API for user management"

boolean

Description: Enable yolo/yes mode (auto-approve all actions)Behavior:

Automatically approves all actions without user confirmation
Forces plain text output mode (no interactive UI)
Exits process automatically when task completes
Useful for automation, CI/CD, and scripting

Example:

cline --yolo "Run the test suite and fix any failures"

Yolo mode bypasses safety confirmations. Only use in controlled environments where you trust Cline to make changes autonomously.

Model Configuration

string

Description: Specify which AI model to use for the taskBehavior: Overrides the configured model for this session only. The setting is not persisted.Supported Models:

Anthropic: claude-sonnet-4-5-20250929, claude-opus-4-20250514
OpenAI: gpt-4o, gpt-4-turbo, o1, o1-mini
Moonshot: kimi-k2.5, kimi-k1.5
Google: gemini-2.0-flash-exp, gemini-1.5-pro
And more (depends on your provider)

Example:

cline --model claude-sonnet-4-5-20250929 "Refactor this module"
cline --model gpt-4o "Write documentation"

boolean | number

Description: Enable extended thinking with a token budgetBehavior:

Without value: Uses default 1024 token budget
With value: Uses specified token budget
Allows the model to “think” more deeply about complex problems
Useful for architecture decisions, complex debugging, and system design

Example:

# Default 1024 tokens
cline --thinking "Architect a microservices system"

# Custom token budget
cline --thinking 2048 "Design a distributed cache"

string

Description: Set the reasoning effort level for the modelValid Values: none, low, medium, high, xhighBehavior: Controls how much computational effort the model puts into reasoning. Higher effort may produce better results but takes longer.Example:

cline --reasoning-effort high "Optimize database queries"
cline --reasoning-effort low "Fix typo in README"

Output Options

boolean

Description: Show verbose output including reasoningBehavior:

Displays detailed reasoning from the AI model
Shows additional debug information
Helpful for understanding decisions and troubleshooting

Example:

cline --verbose "Analyze this codebase architecture"

boolean

Description: Output messages as JSON instead of styled textBehavior:

Forces plain text mode (no Ink UI)
Each message is a JSON object with consistent structure
Suitable for parsing and processing by other tools
Ideal for CI/CD integration and scripting

JSON Structure:

{
  "type": "say",
  "text": "I've analyzed the codebase...",
  "ts": 1704067200000,
  "reasoning": "First, I need to understand...",
  "images": [],
  "files": ["src/index.ts"]
}

Example:

cline --json "What files are in this directory?" | jq '.text'

Directory Options

string

Description: Set the working directory for the taskBehavior: Changes the directory where Cline operates. All file operations are relative to this path.Example:

cline --cwd /path/to/project "Add documentation"
cline --cwd ~/workspace/frontend "Update React components"

string

Description: Path to Cline configuration directoryBehavior: Override the default ~/.cline/data/ directory. Useful for managing multiple configurations or using project-specific settings.Default: ~/.cline/data/Example:

cline --config ~/.config/cline task "Build a feature"
cline --config ./project-cline-config "Run tests"

You can also set the CLINE_DIR environment variable instead of using --config every time.

Safety Options

number

Description: Timeout in seconds for yolo modeDefault: 600 seconds (10 minutes)Behavior: Automatically stops the task after the specified duration in yolo mode. Prevents runaway processes.Example:

cline --yolo --timeout 300 "Quick refactor"

number

Description: Maximum consecutive mistakes before halting in yolo modeBehavior: Safety mechanism that stops execution if Cline makes too many errors in a row. Helps prevent damage in automated workflows.Example:

cline --yolo --max-consecutive-mistakes 3 "Automated testing"

boolean

Description: Reject first completion attempt to force re-verificationBehavior: Forces Cline to review and verify his work before marking a task complete. Improves quality but takes longer.Example:

cline --double-check-completion "Critical production deployment"

Task Management

string

Description: Resume an existing task by IDBehavior:

Loads previous task context and conversation
Optional prompt becomes a follow-up message
Get task IDs from cline history

Example:

# Resume a task
cline --taskId abc123def

# Resume with follow-up
cline --taskId abc123def "Now add unit tests"

# Resume in plan mode
cline --taskId abc123def --plan "What's left to do?"

Environment Variables

CLINE_DIR

Description: Override the default configuration directory Default: ~/.cline/data/ Usage:

export CLINE_DIR=~/.config/cline
cline "Build a feature"

CLINE_COMMAND_PERMISSIONS

Description: Restrict which shell commands Cline can execute Format: JSON object with allow, deny, and allowRedirects fields Fields:

array

Glob patterns for allowed commands. If specified, only matching commands are permitted.Pattern Matching:

* matches any characters
? matches a single character
Setting allow on anything will deny all others

Example:

export CLINE_COMMAND_PERMISSIONS='{"allow": ["npm *", "git *"]}'

array

Glob patterns for denied commands. Deny rules take precedence over allow rules.Example:

export CLINE_COMMAND_PERMISSIONS='{"deny": ["rm -rf *", "sudo *"]}'

boolean

Whether to allow shell redirects (>, >>, <, etc.). Defaults to false.Example:

export CLINE_COMMAND_PERMISSIONS='{"allow": ["cat *"], "allowRedirects": true}'

Rule Evaluation:

Check for dangerous characters (backticks outside quotes, unquoted newlines)
Parse command into segments split by operators (&&, ||, |, ;)
If redirects detected and allowRedirects is not true, command is denied
Each segment is validated against deny rules first, then allow rules
Subshell contents ($(...) and (...)) are recursively validated
All segments must pass for the command to be allowed

Examples:

# Allow only npm and git commands
export CLINE_COMMAND_PERMISSIONS='{"allow": ["npm *", "git *"]}'

# Allow development commands but deny dangerous ones
export CLINE_COMMAND_PERMISSIONS='{"allow": ["npm *", "git *", "node *"], "deny": ["rm -rf *", "sudo *"]}'

# Allow file operations with redirects
export CLINE_COMMAND_PERMISSIONS='{"allow": ["cat *", "echo *"], "allowRedirects": true}'

# Deny only specific dangerous patterns (allow all others)
export CLINE_COMMAND_PERMISSIONS='{"deny": ["rm -rf /*", "sudo rm *", ":(){ :|:& };:"]}'

Configuration Files

Cline stores configuration and state in several JSON files:

globalState.json

Location: ~/.cline/data/globalState.json Contains:

Global settings and preferences
Task history metadata
Mode settings (act/plan)
UI preferences

secrets.json

Location: ~/.cline/data/secrets.json Contains:

API keys and credentials (stored securely)
OAuth tokens
Provider authentication details

Never commit secrets.json to version control or share it publicly.

workspace/

Location: ~/.cline/data/workspace/ Contains:

Workspace-specific state and settings
Per-project configurations

tasks/

Location: ~/.cline/data/tasks/ Contains:

Task history and conversation data
Previous task contexts
Message logs

Debugging

Log Files

Location: ~/.cline/log/ View Logs:

cline dev log

Log Contents:

API request/response logs
Error stack traces
State changes
Command executions

Verbose Mode

Enable verbose output to see detailed reasoning:

cline --verbose "Debug this error"

JSON Mode for Debugging

Use JSON output to inspect message structure:

cline --json "Test command" | jq .

Best Practices

For Development

# Use verbose mode to understand decisions
cline --verbose --thinking "Complex architectural decision"

# Use plan mode for exploratory tasks
cline --plan "Understand this legacy codebase"

# Specify working directory for multi-project workflows
cline --cwd ~/projects/frontend "Update UI components"

For Automation

# Use yolo mode with timeout for safety
cline --yolo --timeout 300 "Run automated tests"

# Use JSON output for parsing
cline --json "Generate report" | jq -r '.text' > report.txt

# Set max consecutive mistakes to prevent damage
cline --yolo --max-consecutive-mistakes 3 "Automated refactoring"

For CI/CD

# Combine with command permissions
export CLINE_COMMAND_PERMISSIONS='{"allow": ["npm *", "git *"]}'
cline --yolo --json "Run tests and fix failures"

# Use specific models for consistency
cline --model claude-sonnet-4-5-20250929 --yolo "Update documentation"

# Set working directory explicitly
cline --cwd $CI_PROJECT_DIR --yolo "Automated code review"

For Production

# Use double-check completion for critical tasks
cline --double-check-completion "Deploy to production"

# Use higher reasoning effort for important decisions
cline --reasoning-effort high "Security audit"

# Review plans before execution
cline --plan "Database migration strategy"
# Then follow up with:
cline --act "Execute the migration plan"

Extension API

CLI Reference

LLM Providers

Global Options

Mode Options

Model Configuration

Output Options

Directory Options

Safety Options

Task Management

Environment Variables

CLINE_DIR

CLINE_COMMAND_PERMISSIONS

Configuration Files

globalState.json

secrets.json

workspace/

tasks/

Debugging

Log Files

Verbose Mode

JSON Mode for Debugging

Best Practices

For Development

For Automation

For CI/CD

For Production

​Global Options

​Mode Options

​Model Configuration

​Output Options

​Directory Options

​Safety Options

​Task Management

​Environment Variables

​CLINE_DIR

​CLINE_COMMAND_PERMISSIONS

​Configuration Files

​globalState.json

​secrets.json

​workspace/

​tasks/

​Debugging

​Log Files

​Verbose Mode

​JSON Mode for Debugging

​Best Practices

​For Development

​For Automation

​For CI/CD

​For Production

Global Options

Mode Options

Model Configuration

Output Options

Directory Options

Safety Options

Task Management

Environment Variables

CLINE_DIR

CLINE_COMMAND_PERMISSIONS

Configuration Files

globalState.json

secrets.json

workspace/

tasks/

Debugging

Log Files

Verbose Mode

JSON Mode for Debugging

Best Practices

For Development

For Automation

For CI/CD

For Production