Anthropic Ships /goal Command for Claude Code: Long-Horizon Objectives
Anthropic shipped a native /goal command for Claude Code — enabling users to define long-horizon objectives with verifiable stopping conditions, setting a declared end state for multi-step autonomous work. The feature shipped mid-week while Nate Herk was recording a 100-hour Claude Code vs. Codex comparison, in direct response to Codex's recently announced experimental /goal mode. The feature-parity treadmill between the two platforms — where exclusive features are matched within days — is now documented in real time across public benchmarking sessions.
Why It Matters
Goal-level objectives with verifiable stopping conditions represent a qualitative shift from task-by-task execution toward declared-intent autonomous operation. The rapid feature matching between Claude Code and Codex confirms both platforms are treating long-horizon autonomy as a primary strategic capability.