Claude Code tips from its creator boost productivity for subscribers

At a glance:

• Slash commands like /btw, /loop and /schedule let you run hands‑off workflows in Claude Code.
• The Claude in Chrome extension gives the model real‑time visual feedback on its own output.
• Mobile, /teleport and Dispatch make Claude Code truly portable across phone, browser and desktop.

What Claude Code is and why slash commands matter

Claude Code is Anthropic’s conversational coding assistant that lets users describe a software task in plain English and receive a working prototype in seconds. While the surface experience feels like chatting with a helpful developer, the real power lies in a set of slash‑prefixed shortcuts that control the session’s behavior. Typing a single forward slash brings up a list of all available commands, and common shortcuts include:

• /model – switch the underlying model version.
• /clear – reset the conversation history.
• /cost – display token usage statistics.
• /btw – ask a quick side question without breaking the current task.
• /loop – run a prompt automatically at a set interval while the session remains open.
• /schedule – create persistent tasks that survive session closure, either on the desktop or in the cloud.
• /teleport – move a cloud session between devices.
• /remote-control – control a local session from a phone or browser.
• --add-dir flag – grant Claude access to additional folders beyond the launch directory.

These commands turn a simple chat interface into a programmable workflow engine, allowing developers to automate repetitive prompts, monitor integrations, and keep Claude’s context fresh without manual copy‑pasting.

The /btw command: asking side questions without interruption

Introduced in March 2026, the /btw (by the way) command lets you pose a quick ancillary query while Claude is mid‑task. The answer appears in a transient overlay, preserving the main conversation flow. Because the side question has full visibility into the current session, you can reference earlier files or outputs, something a parallel session cannot do. For example, you might ask “what was that file you looked at earlier?” and receive an immediate answer, keeping the primary task uninterrupted.

Boris Cherny, the creator of Claude Code, recommends using /btw instead of opening a new session for ad‑hoc queries. The command reduces context switching and keeps the model’s mental model of the project intact, which leads to more accurate follow‑up suggestions.

Seeing what Claude builds with the Chrome extension

One limitation of a pure text‑only interface is that Claude cannot visually verify the code it generates. The Claude in Chrome extension, currently in beta for any paid Claude plan, bridges that gap by giving the model a live browser instance. When Claude suggests HTML, CSS, or JavaScript, the extension renders the output, allowing Claude to “see” the result, check the console for errors, and iterate directly.

In practice, the author used the extension to build an Instagram‑story‑viewer extension. After several failed attempts describing the bug in plain text, enabling the Chrome extension let Claude inspect the rendered UI, spot the missing element, and correct the code automatically. Cherny likens this to giving an engineer a browser to test their own work – a crucial step for reliable, autonomous coding.

Automating repetitive work with /loop and /schedule

Many developers run the same prompts repeatedly, such as checking a mailbox or scanning a Slack channel. The /loop command automates this by re‑executing a prompt at a defined interval while the session stays open. You can specify an exact cadence (e.g., every 30 minutes) or let Claude infer the timing. The loop ceases when the terminal is closed.

For longer‑term automation, /schedule creates tasks that persist beyond the active session. Two flavors exist:

1. Desktop tasks – run on your local machine as long as the Claude app remains open.
2. Cloud tasks – run on Anthropic’s servers, continuing even when your device is off.

Cherny runs multiple loops in parallel for code‑review automation, PR creation from Slack feedback, and stale‑PR cleanup. He encourages users to experiment with turning frequent workflows into “skills” that Claude can execute autonomously.

Managing project scope with the --add-dir flag

When Claude Code starts, it asks permission to access the directory from which it was launched. By default, it cannot see other folders, which can be cumbersome when a project spans several repositories. Using the --add-dir flag when launching Claude expands its view instantly. For example:

claude --add-dir ~/Projects/old-extension

Now Claude can reference both the current project and the legacy extension folder without additional prompts. Adding recurring directories to settings.json makes this behavior permanent for users who regularly work across the same set of folders.

Portability: coding from phone, browser or terminal

Claude Code is not tethered to a single device. Three avenues make it truly mobile:

• Mobile app – Available on iOS and Android, the app includes a “Code” tab where users can start or continue sessions on the go.
• /teleport command – Initiate a cloud session on one device, then pull it into another (e.g., from a laptop to a phone) without losing context.
• Dispatch feature – Turns a phone into a remote control for the Claude Desktop app, letting you manage Slack, files, and email from anywhere.

Cherny notes that he frequently writes code from the mobile app and uses Dispatch daily to stay productive while away from his desk.

The impact of Cherny’s tips on real‑world workflows

Since adopting Cherny’s recommendations, the article’s author reports noticeably faster iteration cycles, fewer back‑and‑forth clarifications, and higher confidence that Claude’s output actually works. The combination of slash commands, visual feedback via the Chrome extension, and cross‑device flexibility turns Claude Code from a novelty chatbot into a practical development partner.

For anyone on a paid Claude subscription, integrating these tips can shave minutes—or even hours—off routine coding tasks, making the token cost of each session more worthwhile.