[Release] anthropics/claude-code v2.1.207: v2.1.207

Claude Code v2.1.207 introduces auto mode availability without opt-in on Bedrock, Vertex AI, and Foundry, while fixing critical issues including terminal freezing during long streaming responses, security consent dialog bypasses, and various UI/workflow bugs. The release improves agent view functionality, Remote Control session handling, and Deep Research labeling, plus enhances security by rejecting shell-injection patterns in plugin hooks and preventing unintended settings persistence.

Key Points

• •Auto mode now enabled by default on Bedrock, Vertex AI, and Foundry (disable via `disableAutoMode` setting)
• •Fixed terminal freezing and keystroke lag when streaming very long lists, tables, paragraphs, or code blocks
• •Fixed security issue where remote managed settings were recorded as consented without showing the consent dialog
• •Eliminated spurious prompt-injection warnings from benign system-generated conversation updates
• •Fixed auto-updater overwriting custom launcher scripts/symlinks at `~/.local/bin/claude` on each release
• •Fixed compound commands with `cd` and `/dev/null` redirects incorrectly prompting for permission
• •Improved agent view: pasting duplicate text now expands collapsed placeholders instead of duplicating entries
• •Fixed Deep Research runs mislabeling Fetch-phase agents as 'unknown'; now displays source hostname
• •Security enhancement: rejected shell-injection patterns (`${user_config.*}`) in plugin hooks; use exec form or `$CLAUDE_PLUGIN_OPTION_<KEY>`
• •Fixed indefinite hang on Windows when AWS credential resolution stalls; 60-second guard now enforces timeout

Workflow Diagram

Concepts