Optimus Plugin Brings Agents Desktop Control
Optimus is an open-source Mercury Agent plugin that lets AI agents see screens, move pointers, type, open apps, and operate desktop software without dedicated APIs across macOS, Windows, and Linux. It launched with Mercury 1.4 and supports approval gates for consequential actions.
Optimus makes computer-use agents more practical by treating the desktop as a universal compatibility layer, though pixel-based control remains less reliable than structured integrations.
- –Screen, mouse, and keyboard control can reach legacy and native apps that expose no API
- –macOS requires Accessibility and Screen Recording permissions; Windows blocks elevated apps
- –Linux support is strongest on X11, while Wayland remains limited
- –Consequential actions such as sending, paying, deleting, or installing require approval
- –Local, cross-platform operation gives developers a compelling self-hosted alternative to cloud-only computer-use agents
DISCOVERED
1h ago
2026-10-25
PUBLISHED
1h ago
2026-10-25
RELEVANCE
AUTHOR
GithubProjects