Slash Commands in Claude Code | CampusX

Slash commands in Claude Code are fast shortcuts typed inside a working session that trigger predefined actions and workflows—often without writing a full prompt. The practical payoff is less repetition: instead of re-explaining the same intent every time, developers can invoke reusable patterns with a short command, making Claude Code feel more like a programming assistant that supports repeatable development flows.

Claude Code supports two kinds of slash commands. Built-in commands ship with the tool and appear by default after installation. Custom slash commands can be created by users to match recurring project-specific workflows—useful when the same requirement keeps repeating across a particular kind of work.

A major portion of the walkthrough focuses on session-related commands, because sessions are the unit of work in Claude Code: each session is one conversation, starts when the user enters the Claude keyword in the command prompt, and ends when the user types /exit. Every session carries a unique ID and stores the full message history, including user messages, assistant responses, and tool messages. Sessions are automatically saved and can be resumed later even after closing the terminal.

Resuming is handled through the /resume flow. The user can resume a prior conversation by using Claude with a resume flag (described as “Claude -r”), which lists past conversations with timestamps. There’s also a /resume slash command that jumps between sessions so work can continue in the middle of switching contexts.

To keep projects organized, the walkthrough recommends a disciplined session strategy: plan and solve one task per session (for example, one session per website feature), close the session when that feature is done, and start a new session for the next feature to avoid mixing contexts. It also advises renaming sessions immediately after starting so the AI doesn’t generate unhelpful names based on early questions. The /rename command lets users set a clear label (e.g., “intro session”). When meaningful milestones are reached, frequent commits are encouraged to capture progress.

Beyond sessions, the walkthrough highlights /by the way (BTW) for side questions that shouldn’t pollute the main conversation history. This is positioned as a way to ask quick reference questions (like what Flask templating is in Python) while Claude continues the primary task; the side answer can be removed from the visible history using a space interaction.

For larger refactors, /export is recommended to save important past conversations into a Markdown file inside the project directory, so the content can be re-provided as context during code changes.

The guide also covers account and model management. /logout and login flows support switching between Claude Code accounts (including Anthropic console-based access). Model choice is framed as a cost/quality tradeoff: Opus for complex planning, Sonnet as the default balance, and HQ for fast, cheap repetitive tasks. A /model command enables switching models on demand, while /usage shows token consumption for the current session and weekly limits. If limits are reached, /extra usage supports midstream top-ups.

Finally, the walkthrough introduces analytics and control commands: /stats provides usage statistics, /insights generates a detailed HTML report after enough sessions to identify better usage patterns, and /config allows changing settings like thinking mode, verbosity, progress bar behavior, and language. /permissions governs tool access by allowing, asking, or denying tool usage (including adding rules for tools like web search or bash commands). The session ends with UI customization via /theme and hands-free prompting via /voice, plus a reminder that typing / and scrolling reveals the full command list.