Claude-Code v2.1.47 Release: Key Fixes and Improvements

The latest Claude-Code release, version 2.1.47, includes several crucial updates and fixes primarily focused on file handling, performance, and compatibility issues, especially on Windows platforms.
Key Updates and Fixes
- File Handling: The
FileWriteToolnow correctly retains intentional trailing blank lines, addressing an issue where they were previously stripped bytrimEnd(). - Windows Terminal Rendering: Bug fixes ensure correct line counts in terminal displays, which were previously affected by
os.EOL(\r\n) inconsistencies, causing misleading line count displays. - Markdown Output: This update resolves the reflow issue where bold and colored markdown text displayed incorrectly on Windows due to line-ending anomalies.
- Memory and Performance Enhancements: Changes have been made to optimize memory usage by releasing API stream buffers, agent context, and skill state post-use, and improving startup performance by deferring
SessionStarthook execution, reducing time-to-interactive by roughly 500ms. - Bash Tool Output: Previous silent discards of bash tool outputs on Windows using MSYS2 or Cygwin shells have been fixed, ensuring outputs are captured and processed as expected.
- Agent Sessions: Improved memory management during long-running sessions by eliminating
O(n²)message accumulation and enabling dynamic indexing and caching of file mentions, which results in performance improvements. - File Mutation Resilience: Single file write/edit errors will no longer abort all other parallel file operations. Independent file mutations are now reliably completed even if one file write/edit operation fails.
- Enhanced Command Capabilities: Configurable multi-line input via the
chat:newlinekeybinding action has been introduced for handling extended commands effectively.
This update is highly relevant for developers working within Windows environments who require reliable terminal and file handling performance. Memory management improvements are advantageous for those leveraging long-running tasks or managing extensive file resources.
📖 Read the full source: GitHub Claude-Code
👀 See Also

Claude Code Subagents Don't Load Skills in Multi-Agent Systems
A developer reports that subagents in Claude Code v2.1.91 cannot access skills defined in .claude/skills/ directory, despite skills working perfectly in the main session. Multiple approaches including skills in agent frontmatter, Skill tool, CLI flags, and Agent Teams all fail.

NVIDIA Vera CPU Launched for Agentic AI Workloads
NVIDIA has launched the Vera CPU, a processor designed specifically for agentic AI and reinforcement learning workloads, claiming 50% faster performance and twice the efficiency compared to traditional rack-scale CPUs.

Agent Monetization Methods Tested: Fastest Result in 80 Seconds
OpenClaw reporters tested multiple agent monetization methods including self-sovereign wallets, prediction markets, DeFi yield farming, bounty hunting, and micropayments. The fastest result was 80 seconds from nothing to a funded Nano wallet via MCP with no API keys, SDK, or human setup.
Public Backlash Against AI Is Real: Violence, Polling Data, and Diminishing Returns
A Molotov attack on OpenAI's CEO, Gen Z anger rising to 31%, and 80% of companies seeing zero productivity gain — the honeymoon is over for AI.