Grok Build July 20: xAI Open-Sources the Full CLI - 844,530 Lines of Rust Under Apache 2.0 - After Repo-Upload Privacy Backlash, Grok Automations Launches Free Scheduled Tasks

Grok Build goes open source: 844,530 lines of Rust under Apache 2.0, published after the repo-upload privacy controversy

• On July 14, researchers found Grok Build had been uploading users' entire directories - Git histories, SSH keys, API keys, cloud tokens, password databases - to xAI's cloud buckets; xAI disabled retention by default and says it deleted the retained data.
• On July 15, xAI open-sourced the full CLI at github.com/xai-org/grok-build (~844,530 lines of Rust, Apache 2.0, a single commit with no history) and reset usage limits; the tree adapts tool code from Codex and OpenCode, still contains the now-disabled upload code, and compiles locally against your own inference via config.toml.

Editor’s read: Open-sourcing the harness is the right response, but the single-commit publish means you must audit the code as-is - there is no history to review.

Grok Automations launch: scheduled tasks free for all users, email triggers require SuperGrok at $30/month

• On July 16, xAI launched Grok Automations: describe a recurring task in plain language and set a schedule (one-time, daily, weekday, weekly, monthly, yearly) free for all Grok users, while email-triggered automations - Grok watches your inbox and fires on sender/recipient/subject matches - require a SuperGrok subscription at $30/month or $300/year.
• Automations integrate with Slack, Google Workspace, GitHub, Airtable and Salesforce connectors plus reusable Skills packages, with full run history, live now at grok.com and in the iOS and Android apps.

Claude Code vs Codex vs Cursor vs Grok Build: the July 2026 state of play for AI coding agents

• On July 19, claude-world.com published a market snapshot noting Grok Build's beta offers terminal-first parallel work, worktrees and.
• as of July 15.
• a fully inspectable open-source harness architecture, and arguing the right tool depends on matching model-plus-harness to workload, failure tolerance and scale.

Cursor doubles included Grok 4.5 usage as Claude Code, Codex and Cursor all reset limits on July 16

• On July 16, Cursor doubled the included usage allowance for Grok 4.5 (alongside Composer 2.5) across all paid tiers, the same day Anthropic and OpenAI refreshed their own quotas.
• Grok 4.5 demand inside third-party tools keeps growing.

Kimi K3: Moonshot AI's 2.8 trillion parameter model becomes the largest open-source release, open weights due July 27

• On July 18, coverage of Moonshot AI's Kimi K3 highlighted a 2.8 trillion parameter open-weight model with a one-million-token context window and a 76% win rate on the Frontend Code Arena.
• a direct rival in the open-source lane Grok Build just entered, with open weights promised by July 27, 2026.