Claude Code command-line flags

Add additional working directories for Claude to read and edit files.

Enable the server-side advisor tool for this session with a model alias or full model ID.

fable, opus, sonnet, or a full model ID

Specify an agent for the current session (overrides the agent setting)

Define custom subagents dynamically via JSON.

Add bypassPermissions to the Shift+Tab mode cycle without starting in it.

Tools that execute without prompting for permission, apart from reads from network paths.

Append custom text to the end of every subagent's system prompt. Only applies in non-interactive mode with -p.

Load text from a file and append it to subagent system prompts.

Append custom text to the end of the default system prompt

Load additional system prompt text from a file and append to the default prompt

Set the auto-compact window for this session without changing your saved settings.

Render screen-reader friendly output: flat text without decorative borders or animations.

Minimal mode: skip auto-discovery of hooks, skills, custom commands, subagents, plugins, MCP servers, auto memory, and CLAUDE.md so scripted calls start faster.

Beta headers to include in API requests (API key users only)

Start the session as a background agent and return immediately.

(Research preview) MCP servers whose channel notifications Claude should listen for in this session.

Space-separated plugin:<name>@<marketplace> entries

Enable Chrome browser integration for web automation and testing

With a task description, create a new cloud session. With a session ID or claude.ai/code URL, queue a message into that session instead, with -p.

Load the most recent conversation in the current directory.

Enable channels that are not on the approved allowlist, for local development.

plugin:<name>@<marketplace> and server:<name> entries

Skip permission prompts. Equivalent to --permission-mode bypassPermissions.

Enable debug mode with optional category filtering, such as --debug='mcp,startup' or --debug='!mcp'.

Write debug logs to a specific file path. Implicitly enables debug mode.

Open the Claude Desktop app on the current directory and exit without starting a session in the terminal.

Disable all skills and commands for this session

Deny rules. A bare tool name removes the matching tools from Claude's context; a scoped rule denies only matching calls.

Set the effort level for the current session. Available levels depend on the model.

low, medium, high, xhigh, max, or ultracode

Removed in v2.1.111. Auto mode is now in the Shift+Tab cycle by default; use --permission-mode auto to start in it.

Create a new cloud session that runs on the self-hosted environment with the given ID. Environment IDs start with ccpool_.

Move per-user context, such as the auto memory location, out of the system prompt and into the first user message.

Run a shell command as a PTY-backed background job instead of starting a Claude session. Use with --bg.

Enable automatic fallback to the specified model(s) when the primary model is overloaded or not available.

Comma-separated list tried in order

When resuming, create a new session ID instead of reusing the original (use with --resume or --continue)

Emit subagent text and thinking blocks in the output stream as assistant and user messages with parent_tool_use_id set.

Open the session picker filtered to sessions linked to a specific pull request.

PR number, GitHub/GitHub Enterprise PR URL, GitLab MR URL, or Bitbucket PR URL

Automatically connect to IDE on startup if exactly one valid IDE is available

Run Setup hooks with the init matcher before the session (print mode only)

Run Setup and SessionStart hooks, then exit without starting a conversation

Include hook lifecycle events in the output stream.

Include partial streaming events in output. Requires --print and --output-format stream-json.

Specify input format for print mode.

text, stream-json

Get validated JSON output matching a JSON Schema after the agent completes its workflow (print mode only).

Run Setup hooks with the maintenance matcher before the session (print mode only)

Stop the run once estimated spend on API calls reaches this amount (print mode only).

Limit the number of agentic turns (print mode only). Exits with an error when the limit is reached.

No limit by default

Load MCP servers from JSON files or strings (space-separated).

Sets the model for the current session with a model alias such as sonnet, opus, haiku, or fable, or a model's full name.

Set a display name for the session, shown in /resume and the terminal title.

Disable Chrome browser integration for this session

Disable session persistence so sessions are not saved to disk and cannot be resumed. Print mode only.

Specify output format for print mode.

text, json, stream-json

Begin in a specified permission mode.

default, acceptEdits, plan, auto, dontAsk, bypassPermissions, or manual (alias for default)

Specify an MCP tool to handle permission prompts in non-interactive mode.

Set who answers permission prompts in print mode.

host (default) or none

Load a plugin from a directory or .zip archive, or several from a folder of plugins, for this session only.

Fetch a plugin .zip archive from a URL for this session only.

Print response without interactive mode.

Emit a prompt_suggestion message with a predicted next user prompt after each turn that generates one.

With --environment, base the new session's checkout on a named ref instead of local HEAD

Deprecated alias for --cloud, including the existing-session form

Start an interactive session with Remote Control enabled so you can also control it from claude.ai or the Claude app.

Prefix for auto-generated Remote Control session names when no explicit name is set.

Defaults to your machine's hostname

Re-emit user messages from stdin back on stdout for acknowledgment.

Start in restricted mode, for evaluation harnesses where Claude Code must not run commands or read the machine's user and project settings.

Resume a specific session by ID or name, or show an interactive picker to choose a session.

Start with all customizations disabled to troubleshoot a broken configuration.

Use a specific session ID for the conversation (must be a valid UUID)

Comma-separated list of setting sources to load.

user, project, local

Path to a settings JSON file or an inline JSON string. Values override the same keys in settings.json files for this session.

Only use MCP servers from --mcp-config, ignoring all other MCP configurations.

Replace the entire system prompt with custom text

Load system prompt from a file, replacing the default prompt

Pass off to rebuild the system prompt on every request instead of reusing the prompt recorded on the conversation's first request.

on (default) or off

Resume a cloud session in your local terminal

Set how agent team teammates display. Overrides the teammateMode setting for this session.

in-process (default), auto, tmux, or iterm2

Create a tmux session for the worktree. Requires --worktree.

--tmux=classic for traditional tmux

Restrict which built-in tools Claude can use.

"" to disable all, "default", or tool names like "Bash,Edit,Read"

Enable verbose logging, shows full turn-by-turn output.

Output the version number

Start Claude in an isolated git worktree at <repo>/.claude/worktrees/<name>.