Claude Code Commands Overview

Overview

This lecture provides an overview of all major commands in Claude Code, explaining their purpose, usage, and best practices for efficient workflow.

Reporting & Troubleshooting Commands

• Use /bug to report bugs directly to the development team.
• /doctor checks if your Claude Code installation has proper permissions and setup.
• /help lists all available commands with brief descriptions.

Managing Context and Token Usage

• /clear wipes the conversation history to start a new chat, reducing token usage.
• /compact summarizes the current chat and starts a new conversation with the summary as context; supports custom instructions.
• Claude Code auto-compacts when at 95% token capacity.

Configuration and Preferences

• /config lets you adjust settings like autocompact, to-do usage, verbose output, theme, notifications, editor mode (normal/vim), and model selection.
• Autocompact is recommended to stay enabled for token management.
• To-do lists keep tasks focused and organized within the tool.
• Verbose output shows Claude Code's internal reasoning; toggle with Ctrl+R when needed.

Project Initialization and Memory

• /init scans your codebase and creates claw.md, summarizing project structure and key components.
• The claw.md file acts as persistent "memory" used at the start of new chats.
• Manual edits to claw.md are allowed for evolving projects.

Login & Session Management

• /login lets you sign in via subscription or API plan.
• /logout logs you out and provides a session summary with costs and models used.
• /cost tracks API usage costs (not relevant for max/pro plan users).

Models and Permissions

• /model selects the Claude model for requests; supports switching between Opus 4, Sonnet 4, and others.
• /permissions manages which commands Claude Code can auto-run (bash, edit/read files, web fetch, MCP tools).
• Be cautious with permissions to avoid unintended file edits or security risks.

Memory Features & Best Practices

• Supports project memory (per-project claw.md) and user memory (preferences across all projects).
• Add custom files or instructions into memory using @filename or the hash # shortcut in chat.
• Organize memory entries in logical groups; be specific and avoid excess entries to keep context clear.

External Integrations

• /mcp connects to external Model Context Protocol servers, allowing Claude Code to interact with external systems (e.g., databases).

GitHub Integration & Status

• /pr_comments fetches pull request comments from GitHub within Claude Code.
• /review triggers a PR review process inside Claude Code.
• /status displays the current working directory, account, active memories, and model.

Editor and Terminal Shortcuts

• /terminal_setup enables Option+Enter for new lines.
• /vim puts the editor in Vim mode for Vim key bindings.

Key Terms & Definitions

• Stateless — The AI model does not retain memory between sessions or chats.
• Token — A unit of text used to measure model usage and compute cost.
• claw.md — A file summarizing project context, used as persistent memory by Claude Code.
• Memory — Stored context (either per-project or user-wide preferences) for Claude Code.
• To-do list — An internal checklist of steps Claude Code creates to keep tasks focused.
• MCP (Model Context Protocol) — A protocol for integrating Claude with external tools and systems.

Action Items / Next Steps

• Run /init when starting with a new project to generate project memory.
• Adjust /config for your preferred workflow and enable autocompact.
• Regularly clear or compact chats to manage token usage.
• Begin adding relevant coding standards and preferences to memory as your workflow evolves.
• Review documented commands with /help as needed.