AI
Claude Code tightens hooks and adds usage warnings
Versions 2.1.299 and 2.1.300 both landed October 14.
Claude Code 2.1.299 adds a setting, CLAUDE_CODE_DISABLE_FILE_MENTIONS, so @ file mentions pass the file's path for Claude to read instead of pasting its contents into your message. UserPromptSubmit hooks (small scripts that run when you send a prompt) now run before a skill or slash command expands, so a hook that blocks a prompt also stops the shell commands embedded in it. And /usage now warns about max effort, workflows, always-on context and large tool results.
Version 2.1.300 adds /advisor on Claude Platform on AWS and splits OpenTelemetry cache-write counts into 5-minute and 1-hour tokens. Hook updatedInput from PreToolUse and PermissionRequest is now validated: unknown keys are dropped and rejected input denies the call. Claude Tag in Slack gains Slack Connect channel support (off until an Owner or admin enables it). On your desk: re-test any skill hook that rewrites tool input, and glance at /usage before your next long session.
Read the original at Claude Code changelog, versions 2.1.299 and 2.1.300 →