v2.1.90

2025-02-22 · Source: Claude Code Changelog · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Advanced, extended

Summary

Claude Code versions 2.1.90 through 2.0.74 introduce a wide array of new features, performance improvements, and bug fixes across its CLI, VS Code extension, and underlying agent capabilities. Key additions include interactive lessons for Claude Code features, enhanced `PreToolUse` hooks with deferral options, and a `PermissionDenied` hook for auto mode. Performance has been significantly improved through optimized JSON serialization, SSE transport, and parallel session loading. Numerous fixes address issues ranging from infinite loops in rate-limit dialogs and prompt-cache misses to memory leaks, UI crashes, and security vulnerabilities in PowerShell tool permission checks. The updates also expand `CLAUDE.md` and skill loading, refine `--resume` behavior, and introduce new environment variables for customization and debugging.

Key takeaway

For AI Architects and NLP Engineers managing Claude Code deployments, these updates significantly improve stability, performance, and security. Your teams should review the new hook capabilities, especially `PermissionDenied` and `PreToolUse` deferral, to enhance custom workflows and integrate advanced permission handling. Additionally, the performance improvements in large sessions and tool interactions mean more efficient resource utilization, reducing operational costs and improving developer experience. Consider updating to the latest version to benefit from these enhancements and address critical bug fixes.

Key insights

Recent Claude Code updates enhance performance, fix critical bugs, and expand customization and agent interaction capabilities.

Principles

• Prioritize security hardening in tool interactions.
• Optimize performance for large-scale operations.
• Provide granular control over agent behavior.

Method

The updates involve optimizing data structures, parallelizing I/O operations, refining permission checks, and expanding hook functionality to improve stability, performance, and user control.

In practice

• Use `/powerup` for interactive Claude Code lessons.
• Configure `PreToolUse` hooks for custom tool call logic.
• Leverage `--resume` for faster project session loading.

Topics

• Plugin System
• Agent Teams
• Remote Control
• Performance Optimization
• Security & Permissions

Code references

• anthropics/claude-code

Best for: AI Architect, NLP Engineer, CTO, AI Engineer, Machine Learning Engineer, Software Engineer

Related on AIssential

• v2.1.63 — Recent Claude Code updates prioritize memory efficiency, multi-agent capabilities, and robust cross-platform performance.
• v2.1.32 — Recent Claude Code updates enhance performance, fix critical bugs, and expand agent capabilities for developers.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Claude Code Changelog.