Claude-Code v2.1.41 Release: Key Updates and Fixes
The latest Claude-Code release v2.1.41 mainly focuses on bug fixes, efficiency improvements, and added features for better usability. Notably, it addresses AWS authentication issues by implementing a 3-minute timeout to prevent indefinite hanging. Additionally, new CLI subcommands like claude auth login, claude auth status, and claude auth logout have been added for improved authentication handling.
Windows ARM64 (win32-arm64) now has native binary support, enhancing the software's cross-platform compatibility. The /rename command has been improved to auto-generate session names based on conversation context when no arguments are provided, and the terminal layout for the prompt footer has been optimized for narrow setups.
Several fixes have been implemented, including resolving file resolution issues with @-mentions that contain anchor fragments (e.g., @README.md#installation). The FileReadTool has been patched to avoid blocking on FIFOs, /dev/stdin, and large file sizes. Background task notifications are now correctly delivered in the streaming Agent SDK mode.
Other notable fixes address UI and functionality glitches: preventing the cursor from jumping on each keystroke in classifier rule inputs, ensuring markdown link display text is preserved, fixing auto-compact failure error notifications, correcting elapsed time displays by excluding permission wait times, and clearing stale permission rules when settings are updated on disk. Additionally, hook blocking errors that used to show stderr content in UI have been addressed.
📖 Read the full source: GitHub Claude-Code
👀 See Also
Claude vs GPT-4o: Same Double Pendulum Prompt, Different Coordinate Conventions
Claude and GPT-4o produce visually different double pendulum simulations because they interpret theta from opposite verticals — top vs bottom — while using the same renderer. The math is correct in both cases, but the mismatch reveals a subtle ambiguity in prompt interpretation.
Cambridge researchers develop hafnium oxide memristor for low-energy AI chips
Researchers at the University of Cambridge have created a hafnium oxide-based memristor that switches currents a million times lower than conventional oxide devices, potentially reducing AI hardware energy use by up to 70%.
OpenClaw Agent System Broken After Recent Updates
Recent OpenClaw updates have broken core agent functionality, with users reporting that agents can't be reliably created or run. The system previously allowed creating agents, having them appear correctly, running workflows, and using them for real tasks.
Reddit post discusses internal repair loops for no-code creative AI
A Reddit post argues that no-code creative AI systems need internal repair mechanisms to handle common-sense failures like impossible mechanical structures or distorted anatomy, rather than making users debug outputs.