New Features #
Configurable Output Token Limit
- New environment variable:
CLAUDE_CODE_MAX_OUTPUT_TOKENS - Allows customization of the maximum output tokens for Claude's responses
- Default remains 32,000 tokens for most models
- Automatically set to 8,192 for Claude 3.5 and Haiku models
- Example usage:
export CLAUDE_CODE_MAX_OUTPUT_TOKENS=50000
Enhanced Telemetry and Debugging
- New environment variable:
OTEL_LOG_USER_PROMPTS - Enables logging of user prompts for debugging purposes
- Example usage:
export OTEL_LOG_USER_PROMPTS=1