COMMAND REFERENCE▌

Manage agent configurations and monitor background agents running in parallel.

Detach the current session to run as a background agent and free this terminal. Optionally pass a final instruction before detaching.

Orchestrate large-scale changes across a codebase in parallel. Decomposes work into 5–30 independent units, each in its own git worktree with its own PR.

Ask a quick side question without adding it to the main conversation history.

Start a new conversation with empty context. The previous conversation stays available in /resume. Optionally name the previous conversation.

Review the current diff for correctness bugs and for reuse, simplification, and efficiency cleanups. Pass --fix to apply findings, --comment to post as GitHub PR comments, or ultra for a deep cloud review.

Free up context by summarizing the conversation so far. Optionally pass focus instructions for the summary.

Visualize current context usage as a colored grid. Shows optimization suggestions for context-heavy tools and capacity warnings.

Open an interactive diff viewer showing uncommitted changes and per-turn diffs. Navigate with arrow keys between the git diff and individual Claude turns.

Diagnose and verify your Claude Code installation and settings. Press f to have Claude fix any reported issues.

Set the model effort level: low, medium, high, xhigh, max, or ultracode. Takes effect immediately without waiting for the current response.

Spawn a forked subagent that inherits the full conversation and works on a directive while you keep going in the main session.

Set a persistent goal — Claude keeps working across turns until the condition is met. View, pause, resume, or clear with subcommands.

View hook configurations for tool events — pre- and post-tool hooks that run shell commands around Claude's actions.

Initialize project with a CLAUDE.md guide covering repo conventions, commands, and agent instructions.

Run a prompt repeatedly on a schedule while the session stays open. Omit the interval and Claude self-paces between iterations.

Manage MCP server connections and OAuth authentication. Browse servers, reconnect disconnected ones, or enable/disable them.

Edit CLAUDE.md memory files, enable or disable auto-memory, and view auto-memory entries.

Switch the AI model and save it as your default for new sessions. Adjust effort level with left/right arrows on models that support it.

Manage allow, ask, and deny rules for tool permissions. Opens a dialog to view rules by scope, add or remove rules, and review auto-mode denials.

Enter plan mode to design your approach before coding. Claude proposes an execution plan for review before any implementation starts.

Resume a previous conversation by ID or name, or open the session picker.

Rewind the conversation and/or code to a previous checkpoint, or summarize from a selected message.

Analyze pending branch for security vulnerabilities — injection, auth issues, data exposure, and other OWASP risks.

Review changed code for cleanup opportunities and apply the fixes. Four review agents run in parallel covering reuse, simplification, efficiency, and abstraction level.

List available skills and manage which are shown to Claude. Press t to sort by token count, Space to hide a skill.

View and manage everything running in the background in the current session.

Show session cost, plan usage limits, and activity stats. Includes breakdown by skill, subagent, plugin, and MCP server on paid plans.

Approve one retry of a recent auto review denial — lets Codex attempt a blocked action once under current policy.

Archive the current session and exit. Removes it from active session lists; transcript kept locally. Restore with 'codex unarchive <SESSION>'.