Question 1

What is the /btw command in Claude Code and why is it useful?

Accepted Answer

The /btw command lets me ask a quick question about what I’m doing without adding it to the main conversation history. That means I can get a fast answer without cluttering context or pulling Claude off a long-running task. It’s basically multitasking inside the same Claude Code instance while the main job keeps running.

Question 2

How does the /loop command work in Claude Code?

Accepted Answer

/loop is like a session-level cron for prompts. I can specify an interval and a prompt, and Claude Code will run it repeatedly on that schedule. It’s perfect for polling status updates, generating recurring summaries (like recently merged PRs), or even setting one-time reminders.

Question 3

What’s the difference between /loop and Claude Code Desktop Scheduled Tasks?

Accepted Answer

I think of /loop as temporary scheduling inside a running session, while Desktop Scheduled Tasks are persistent local automation. Scheduled tasks can start a new local session at a time and frequency I choose, as long as my computer is awake and the desktop app is open. That makes them ideal for recurring workflows like daily code reviews or dependency updates.

Question 4

Can Claude Code send scheduled task outputs to Telegram?

Accepted Answer

Yes—this is one of the coolest practical workflows. I can have Claude add a Telegram messaging module to my repo, create a bot (like via BotFather), and put the credentials in my .env. Then I end my scheduled-task prompt by telling Claude to send the output to Telegram, and it will deliver the result every time it runs.

Question 5

What changed with Claude Code memory writing rules in the latest update?

Accepted Answer

The system prompt now treats the automemory directory as already existing, so Claude should write directly using the right tool instead of running mkdir or checking paths. On top of that, memories are now enforced into a structured format: the rule/fact, why it matters, and how it should influence behavior. It’s a subtle change, but it makes context retention across sessions way better.

Question 6

What is Agent Skills 2.0 and what’s new about it?

Accepted Answer

Agent Skills creator got revamped so authors can test, measure, and refine skills without coding. It’s designed to catch regressions and make sure skills trigger correctly, and it even supports multi-agent parallel testing. In the video I mention an example where a PDF-filling skill was fixed by anchoring text positioning to extracted coordinates for consistent output.

Question 7

Does Claude Code have code review and who can use it?

Accepted Answer

Claude Code introduced a deep multi-agent code review system that can catch bugs even humans miss. The downside is it’s expensive—around $15–$25 per run—and it’s currently only for Team and Enterprise users. It doesn’t approve PRs, but it closes the review gap and helps teams move faster.

Question 8

What are effort levels in Claude Code and why do they matter?

Accepted Answer

Now when I start a session, Claude Code asks what effort level I want: low, medium, high, max (and there’s also ultra mode I’ve talked about). This controls how deeply Claude reasons, how long it works, and how much it costs. Setting it upfront gives me more predictable results and token usage.

Claude Code 2.0 MASSIVE Upgrade! (Game Changer)

🛍️ Products Mentioned (8)

Agent Skills 2.0

loop

Scheduled Tasks

Become a Patron (Private Discord)

Learn to code with Scrimba – from fullstack to AI

Intheworldofai Product

Discord Product

Skool Product

About This Video

Frequently Asked Questions

🎬 More from WorldofAI