Claude Code 3.0: New Features and Usage Limit Adjustments

Major Feature Upgrades

Claude Code has received significant updates focused on multi-agent orchestration and developer workflow integration. The primary model is now Claude Opus 5, which supports a 1-million-token context window on eligible plans. A new /design command allows users to generate editable UI artboards, while cross-session messaging enables separate instances to coordinate tasks by referencing each other via the @ symbol. Fork mode is now enabled by default, allowing the system to spawn background sub-agents that inherit the current conversation context and prompt cache, with support for up to 20 concurrent sub-agents.

Security and Automation

Anthropic introduced a dedicated security plugin that utilizes multi-agent systems to map codebase architecture, build threat models, and identify vulnerabilities. The plugin can scan branches, diffs, or specific commits and generate patches for review. Additionally, Auto mode is now the default permission setting for new sessions on Pro, Max, and Team plans. A background classifier handles routine actions, and these classification calls no longer count toward user usage limits.

Usage Limit Adjustments

Anthropic announced a permanent 25% increase to the original baseline weekly usage limits, effective September 14. However, because this replaces a temporary 50% boost that was in effect for several months, users will experience a 17% reduction in available capacity compared to the limits provided during the temporary period. To mitigate the impact of these caps, the system now includes an auto-continue feature that resumes sessions automatically once limits reset, and heavy users can opt to pay for additional usage via API rates.