---
title: "12 Claude Code updates that shipped this week and what each one does"
description: "5 source-backed AI claims from 12 Claude Code updates that shipped this week and what each one does, with key statistics, context, and the original AI…"
url: "https://kbanc.com/claims-library/12-claude-code-updates-that-shipped-this-week-and-what-each-one-does"
source: "https://aiadopters.club/p/12-claude-code-updates-that-shipped"
date: "2026-03-13"
topics: ["tools", "implementation"]
generated: "2026-08-31"
---

# 12 Claude Code updates that shipped this week and what each one does

By Kamil Banc | March 13, 2026

## Claims

1. **Dismissible /btw Answers** (source summary): The /btw command answers questions in a dismissible overlay without polluting conversation history.
2. **Scheduled /loop Prompts** (source summary): The /loop command runs prompts on recurring schedules that expire after 72 hours.
3. **Structured Auto Memory** (source summary): Claude Code's auto memory now uses a structured three-part format for better cross-session recall.
4. **Multi-Agent Code Review** (source summary): Multi-agent code review scans entire codebases for bugs and costs approximately $15 to $25 per run.
5. **Priority for Daily Coders** (Kamil's interpretation): Starting with /btw and /loop is the recommended priority for developers who write code daily.

## Evidence

### Quote
> "Not incremental patches. Structural changes to how the tool thinks, remembers, schedules, and reviews your work." - Kamil Banc

### Key Statistics
- **72 hours**: Maximum lifetime of any /loop scheduled task before it automatically expires
- **$15 to $25**: Approximate cost per run of the multi-agent code review feature, limited to Team and Enterprise subscribers
- **12 updates**: Number of meaningful Claude Code updates Anthropic shipped within the span of days
- **4 effort levels**: Every session now prompts users to choose low, medium, high, or max reasoning effort

## Context
The article is a practitioner-authored changelog walkthrough by Kamil Banc, cataloging twelve Claude Code updates shipped within days and explaining the practical workflow impact of each. Claims are drawn directly from the author's descriptions of the new features, including command syntax examples, guardrails like the 72-hour task expiry, and pricing details for the multi-agent review system. The author closes with role-based prioritization guidance, mapping each update to daily coders, report builders, team managers, and API developers. Readers evaluating these claims should note the piece is promotional-adjacent commentary rather than independent benchmarking, and no external primary sources were supplied to verify the feature behavior. Practitioners can apply the findings by testing two updates, such as /btw and /loop, in real sessions to assess fit.

## Source
- Original: [12 Claude Code updates that shipped this week and what each one does](https://aiadopters.club/p/12-claude-code-updates-that-shipped)
- Cite: kbanc.com/claims-library/12-claude-code-updates-that-shipped-this-week-and-what-each-one-does