Claude Code Update Delivers New Features and Bug Fixes
Anthropic has released a comprehensive update to Claude Code, its AI-powered coding assistant, introducing significant architectural improvements, expanded enterprise controls, and cross-platform stabilizations across terminal, integrated development environments, and cloud workspaces. The update introduces several foundational enhancements for workflow and security. Claude Code now reads AGENTS.md as an alternative configuration file, streamlining project setup. Network and gateway administrators gain finer control through the CLAUDE_GATEWAY_PROXY_IS_EGRESS_BOUNDARY flag and optional upstream headers, ensuring stricter egress routing for cloud deployments. Remote and self-hosted sessions now support fast mode, accelerating feedback loops where organizational policies permit. Artifact management has been overhauled, enabling inline updates, expanded publishing capabilities, and robust conflict resolution across distributed worktrees. Additionally, the platform now synchronizes skills and plugins from claude.ai accounts, with granular opt-out controls for enterprise environments. Enterprise and security governance see substantial upgrades. Managed policy enforcement has been tightened, with stricter validation for MCP server configurations and plugin marketplaces. Sandbox permissions now conduct more precise command analysis, blocking ambiguous shell expansions and improving directory access controls. Telemetry and observability are expanded through new OpenTelemetry trace attributes, enhanced gateway boot diagnostics, and improved error reporting for credential refreshes, proxy failures, and rate limits. Cloud session environments now include dedicated read-only summaries for administrators, with refined network access policies and automated GitHub routine scheduling that gracefully handles token renewals and connection failures. Stability and reliability form the core of this release, addressing hundreds of edge cases that previously disrupted developer workflows. Session resumption, background task management, and prompt caching have been fundamentally rewritten to prevent context loss, duplicate tool calls, and rendering freezes. The VS Code extension receives targeted fixes for plugin installation, MCP server reconnection, and permission dialog persistence, while terminal performance improvements reduce latency during large file diffs and extended reasoning sessions. Platform-specific issues on Windows, Linux, and macOS have been resolved, including fixes for Wayland startup behavior, network path permissions, and shell variable expansion. Slack integration via Claude Tag now features improved channel synchronization, accurate credential routing, and refined notification delivery. This release positions Claude Code as a more resilient and enterprise-ready development tool. By addressing foundational session management, tightening security boundaries, and standardizing error handling across all interfaces, Anthropic has delivered a platform that scales reliably for both individual developers and large engineering organizations. The update is available across all supported channels, with immediate benefits for teams adopting AI-assisted coding in regulated or complex network environments.
