xAI's Grok Build '/goal' lets AI autonomously complete multi-step coding tasks
New '/goal' command packs verification and long-running autonomy into one terminal command.
xAI introduced '/goal' on June 22, 2026, as a new feature within its Grok Build terminal-based coding agent. The mode is designed for long-running, autonomous task execution where the AI agent can independently work through complex implementation tasks. It includes built-in verification steps—like reviewing code, inspecting webpages, or executing scripts—to ensure each stage is completed correctly before moving forward. By packaging this autonomy behind a single '/goal' command, developers can kick off multi-step workflows and monitor progress via a visible checklist, reducing the need for constant manual supervision.
This release marks a significant step in xAI's push toward more self-sufficient coding agents. Unlike traditional interactive coding tools that require step-by-step prompting, '/goal' allows the agent to handle entire sequences of tasks from start to finish, verifying its own work along the way. For developers working on large refactoring, dependency updates, or integration tasks, this can dramatically cut down iteration time. The feature is available now within Grok Build, and early users report that the autonomous verification reduces bugs introduced during complex automated edits.
- Released by xAI on June 22, 2026 as a new mode in Grok Build terminal agent.
- Enables autonomous long-running coding tasks with built-in verification (code review, webpage inspection, script execution).
- Operates via a single '/goal' command and displays a visible progress checklist for multi-step workflows.
Why It Matters
Grok Build's '/goal' reduces manual oversight in complex coding, accelerating autonomous multi-step development workflows.