Antgrid brings the coding agents on all your machines into one desktop app. Every session on every machine, in one list. Turn on remote access on a machine and open its sessions from another, review the agent's changes and commit.
It runs the CLI you already installed: Claude Code, Codex, opencode, Cursor, Copilot, Antigravity, Kilo, Kimi and Mistral Vibe. Each agent loads its own config, instructions, skills, hooks and MCP servers, as it does in your terminal.
Handler takes the follow-ups on long-running work. You tell it how the work should proceed and when to ask you. It checks the agent's replies against those instructions, asks for evidence before it closes a step, and brings you the questions only you can answer. It works in Claude Code, Codex and opencode sessions. Its judge is your own agent CLI on your own provider account: no Antgrid model in the loop.
Traffic between your devices is end-to-end encrypted. When there is no direct route, a relay forwards the encrypted packets without a key to them. Antgrid's servers never see your code, prompts or terminal output.
Also in the app: scheduled prompts that run while the desktop app is open; Claude Code and Codex sessions on the same repository can message each other; isolated sessions, each on its own branch; a file explorer, Git review, browser preview and push notifications.
Windows (Microsoft Store), macOS (Apple Silicon) and Android (Google Play). Free during the beta, with Handler included.
The source is public, and most of it is open source. The app, bridge and wire protocol are MPL-2.0; the relay and account service are ELv2.
antgrid-ai 
No comments or reviews, maybe you want to be first?