Claude Code v2.1.59 adds auto-memory, copy command, and shell improvements
What's new in Claude Code v2.1.59
Anthropics released Claude Code v2.1.59 on February 26, bringing several practical improvements to the AI coding assistant. This release focuses on memory management, code interaction, and shell command handling.
Key features and fixes
- Auto-memory management: Claude now automatically saves useful context to auto-memory. You can manage this with the
/memorycommand. - Interactive copy command: Added
/copycommand that shows an interactive picker when code blocks are present, allowing selection of individual code blocks or the full response. - Improved bash command handling: Enhanced "always allow" prefix suggestions for compound bash commands (e.g.,
cd /tmp && git fetch && git push) to compute smarter per-subcommand prefixes instead of treating the whole command as one. - Task list ordering: Improved ordering of short task lists.
- Memory optimization: Reduced memory usage in multi-agent sessions by releasing completed subagent task state.
- Bug fixes: Fixed MCP OAuth token refresh race condition when running multiple Claude Code instances simultaneously. Fixed shell commands not showing clear error messages when the working directory has been deleted.
This release addresses specific workflow issues developers encounter when using Claude Code for shell operations and code manipulation. The auto-memory feature should help maintain context across longer sessions, while the improved bash command handling makes compound operations more reliable.
📖 Read the full source: GitHub Claude-Code
👀 See Also
Meera: A Fully Offline AI Assistant for Linux Gnome Built on Qwen3.5-2B
Meera is an offline AI assistant for Gnome Desktop that uses Qwen3.5-2B-Q4_K_M (1.2 GB) and llama-cpp with Vulkan support. It leverages a second tiny embedding model for tool selection and RAG, avoiding prompt embedding bloat. Works on Ubuntu 24.04 with RTX 5090 and Fedora Silverblue on Intel i3.
Meta Ads MCP OAuth Works But Most Ad Accounts Not Enabled Yet
Meta Ads MCP OAuth flow works and loads 29 tools, but ads_get_ad_accounts returns is_ads_mcp_enabled: false with a message that the feature is gradually rolling out.
Local 35B MoE Model Drops Agent OS Code Failure Rate to 0%
A developer reports that switching a multi-agent system's runtime to Qwen 3.6 35B A3B (MoE, 3B active params) eliminated code failures, achieving 100% success rate through a 5-layer validation gate.
Shieldbot: Open-Source Security Scanner Plugin for Claude Code
Shieldbot is an open-source security scanner that runs as a plugin inside Claude Code, integrating six scanners including Semgrep with 5,000+ rules, Bandit, Ruff, detect-secrets, pip-audit, and npm audit. It deduplicates findings and generates prioritized reports with risk scores and code fixes.