Grok Build 1.0: xAI's coding-agent CLI leaves beta
xAI's terminal agent goes stable, joining Claude Code and Codex as a daily driver — while GitHub quietly stops auto-assigning Copilot to review your PRs.
Copy markdown
A Grok-native rival to Claude Code and Codex
Grok Build hit 1.0 on Aug 7, leaving beta. Install with one curl line (needs SuperGrok or X Premium Plus); it ships plan-before-execute mode, parallel subagents on git worktrees, headless `-p` scripting, and reads your existing AGENTS.md, MCP servers, hooks, and plugins — so it drops into a Claude Code or Codex setup with no rewiring.
The 1.0 changes you'll feel first
Dashboard rows now summarize what the agent did on its last turn, and permission prompts print the full script before you approve it — fewer blind yes/no's on commands you can't read. Session restore is faster and large directories no longer choke the agent.
GitHub stops auto-assigning Copilot to your PRs
Since Aug 7, GitHub Code Quality no longer auto-creates a ruleset requesting Copilot review on every pull request — after users said choosing a reviewer should be their call. If Copilot reviews quietly vanished from your PRs, that's why; re-enable per repo or org in ruleset settings. Affects Enterprise Cloud and Team.