Overview
- OpenAI released a dedicated macOS Codex app that orchestrates multiple coding agents in parallel using isolated threads and Git worktrees to avoid conflicts.
- The app adds reusable Skills and scheduled Automations for tasks like issue triage, CI failure summaries, release briefs, and bug checks, with results queued for human review.
- OpenAI says the app runs on GPT‑5.2‑Codex, offers a configurable system‑level sandbox with permission approvals, and is temporarily available to Free and Go users, with doubled rate limits for paid tiers.
- Apple’s Xcode 26.3 Release Candidate integrates Anthropic’s Claude Agent and OpenAI’s Codex via the Model Context Protocol, enabling agents to search documentation, analyze projects, adjust settings, and verify work with Xcode Previews.
- Independent tests report mixed standings for GPT‑5.2, topping TerminalBench but not showing a clear edge on SWE‑bench, reflecting the difficulty of benchmarking agentic workflows.