CLI Flags

Command-line flags and options for Happy CLI

Happy CLI supports all Claude Code flags plus additional Happy-specific flags for enhanced functionality.

Happy-Specific Flags

These flags are unique to Happy CLI and control Happy-specific features.

—yolo

Shortcut for --dangerously-skip-permissions (bypasses permission prompts).

happy --yolo

Use with caution. This bypasses all permission checks and allows the AI to execute commands without confirmation.

—chrome

Enable Chrome browser access for this session.

happy --chrome

Allows Claude to interact with your Chrome browser during the session.

—no-chrome

Disable Chrome mode even if it’s enabled by default in settings.

happy --no-chrome

—no-sandbox

Disable Happy’s OS-level sandbox for this session.

happy --no-sandbox

Sandboxing must be configured first using happy sandbox configure. This flag bypasses the sandbox for a single session.

—js-runtime

Specify the JavaScript runtime to use for spawning Claude Code.

happy --js-runtime bun
happy --js-runtime node

Options: node, bun

—claude-env

Pass environment variables to Claude Code.

happy --claude-env ANTHROPIC_BASE_URL=http://127.0.0.1:3456
happy --claude-env CUSTOM_VAR=value

Multiple environment variables can be passed:

happy --claude-env VAR1=value1 --claude-env VAR2=value2

Format: KEY=VALUE

Use --claude-env ANTHROPIC_BASE_URL to route requests through a custom API endpoint like claude-code-router.

—happy-starting-mode

Specify the starting mode for Happy.

happy --happy-starting-mode local
happy --happy-starting-mode remote

Options:

• local: Start in local/terminal mode
• remote: Start in remote/mobile-controlled mode

—started-by

Indicate how the session was started (internal flag).

happy --started-by daemon
happy --started-by terminal

Options: daemon, terminal

This flag is typically used internally by Happy daemon and not by end users.

Common Claude Code Flags

Happy supports all Claude Code flags. Here are the most commonly used:

—help, -h

Show help information.

happy --help
happy -h

—version, -v

Show version information.

happy --version
happy -v

—resume

Resume a previous Claude session.

happy --resume
happy --resume <session-id>

When resuming, a new session is created with the complete history from the original session.

—print

Print output to stdout instead of interactive terminal.

happy --print "list files in this directory"

—output-format

Specify output format when using --print.

happy --print --output-format stream-json "analyze this code"
happy --print --output-format text "what's in this file?"

Options: stream-json, text

—verbose

Enable verbose output.

happy --verbose

—dangerously-skip-permissions

Skip all permission prompts (use with caution).

happy --dangerously-skip-permissions

This is equivalent to --yolo and should be used with extreme caution.

ACP Agent Flags

Flags specific to ACP (Agent Client Protocol) agents like Gemini.

—verbose (ACP)

Print raw ACP backend and envelope events.

happy acp gemini --verbose
happy acp opencode --verbose

— (double dash)

Delimiter for custom ACP commands.

happy acp -- opencode --acp
happy acp -- custom-agent --custom-flag

Everything after -- is passed directly to the ACP agent command.

Subcommand-Specific Flags

Notify Command

happy notify -p "message" [-t "title"]

Flag	Description	Required
-p	Notification message	Yes
-t	Notification title	No

Example:

happy notify -p "Deployment complete!"
happy notify -p "Build finished" -t "CI/CD"

Auth Command

happy auth login [--force]
happy auth logout
happy auth whoami

Flag	Description
--force	Force re-authentication even if already logged in

Daemon Command

happy daemon start
happy daemon stop
happy daemon status
happy daemon list
happy daemon logs

No special flags; each subcommand is self-contained.

Sandbox Command

happy sandbox configure
happy sandbox status
happy sandbox disable

No flags; uses interactive prompts for configuration.

Connect Command

happy connect [provider]

Connect AI vendor API keys. Providers: anthropic, openai, gemini, etc.

Flag Combinations

Flags can be combined for powerful workflows:

Development with Custom API

happy --js-runtime bun --claude-env ANTHROPIC_BASE_URL=http://localhost:3456 --chrome

Safe Testing

happy --no-sandbox --verbose

Automated Scripting

happy --print --output-format stream-json "run tests" | jq

Resume with Chrome

happy --resume <session-id> --chrome

Flag Precedence

When multiple configuration sources exist, flags take precedence in this order:

1. Command-line flags (highest priority)
2. Environment variables
3. Settings file (~/.happy/settings.json)
4. Defaults (lowest priority)

Example:

# settings.json has chromeMode: true
# But --no-chrome flag disables it for this session
happy --no-chrome

Tips and Best Practices

Use --yolo carefully

The --yolo flag bypasses all safety checks. Only use it in trusted environments or for well-understood tasks.

Chrome mode for web tasks

Enable --chrome when you need Claude to interact with web content or test web applications.

Custom API endpoints

Use --claude-env ANTHROPIC_BASE_URL to route through proxies or local development servers:

happy --claude-env ANTHROPIC_BASE_URL=http://127.0.0.1:3456

Resume sessions efficiently

Use --resume without arguments to get a list of recent sessions to choose from.

Combine with Unix tools

Happy’s --print --output-format stream-json works great with jq and other CLI tools:

happy --print --output-format stream-json "analyze code" | jq '.message.content'

All Flags Reference

Session Control

--resume, --print, --output-format, --verbose

Safety

--yolo, --dangerously-skip-permissions, --no-sandbox

Features

--chrome, --no-chrome, --js-runtime

Configuration

--claude-env, --happy-starting-mode, --settings

Getting More Help

For the complete list of Claude Code flags:

happy --help

For subcommand-specific help:

happy <subcommand> --help
happy auth --help
happy daemon --help
happy sandbox --help