Codex lands on Windows with native agent sandbox and v0.110 plugin upgrades
OpenAI's Codex desktop app now runs natively on Windows with a hardened agent sandbox, and the latest v0.110 update brings a plugin system, richer multi-agent flows, and service tier controls. The Windows release delivers a “first Windows-native agent sandbox” with restricted tokens, filesystem ACLs and dedicated sandbox users, plus native PowerShell support, multi‑agent orchestration, centralized code diff review, and broad IDE/editor integration including Visual Studio, Rider/PhpStorm, Git Bash, GitHub Desktop, Cmder, WSL, and Sublime Text ([TechRadar](https://www.techradar.com/pro/openai-releases-a-windows-version-of-codex-coding-app), [OpenAI community post](https://community.openai.com/t/the-codex-app-is-now-on-windows/1375704)). For teams standardizing on Windows dev environments, this removes the WSL/VM dependency while keeping strong OS‑level isolation. Release [v0.110.0](https://github.com/openai/codex/releases/tag/rust-v0.110.0) adds a plugin system for loading skills, MCP entries, and app connectors (with install endpoints), expands multi‑agent TUI flows (approval prompts, /agent enablement, clearer prompts, role‑labeled handoffs), introduces persisted /fast toggles with app‑server “fast/flex” tiers, improves memories with workspace‑scoped writes and guardrails, and ships a direct Windows installer script; it also fixes Windows state DB paths and stability issues. Early community feedback flags an Automations (Beta) test‑run issue that can hang IN_PROGRESS without output ([community bug report](https://community.openai.com/t/possible-bug-codex-automations-beta-test-runs-stay-in-progress-with-no-output/1375778)).