antgrid · mission control for your AI coding agents

Starting an agent is easy. Knowing it finished isn't.

Run Claude Code, Codex and Cursor on your own machine — then watch, steer and approve them from your phone. End-to-end encrypted.

antgrid's supervisor stays with the run: it answers what it can, wakes you for what it can't, and calls a task done only with evidence — test output, exit codes — never the agent's say-so.

Start free Download

Desktop out now · iOS & Android coming to the App Store & Play.

macbook-pro · api watching
Migrate auth to session tokens
done when: auth tests pass and typecheck is clean
2 of 3 follow-ups done evidence
auth suite 142 passed · exit 0
typecheck tsc exit 0 · 0 errors
changelog entry waiting — no diff yet
03:40 woke you — the migration drops a column, needs your call

Fleet view

Every agent. Every machine. One screen.

Your laptop, your workstation, a cloud box — every agent you're running, grouped by the machine it's on, with the ones that need you floated to the top.

Agents · 6 total
1 needs you 4 working 1 done
macbook-pro
api Claude Code Refactoring auth middleware… 2m
web Codex Writing checkout tests… 5m
studio-workstation
relay Claude Code Needs you — which migration strategy? just now
app Cursor Built release bundle 12m
prod-box · cloud
evals Codex Running E2E suite… 1m
infra Claude Code Tailing deploy logs… 3m

End-to-end encrypted

The relay can't read a byte.

Everything you just saw — terminal output, your prompts, the files, every approval — is sealed on your machine and only opens on your phone. The server in the middle just forwards blobs it can never decrypt.

The alternative is letting someone else's machine hold your code. antgrid runs the agent on yours — and the relay in between is built so it can't read what passes through.

What you see decrypted 
studio-workstation · Claude Code
02:14  Editing auth/session.ts
       + verifyPairApproval(sig, uuid)
✓ 142 files typechecked
? approve destructive migration?
What the relay sees ciphertext 
k7Vd2+Rm9pQ1fX0aZ
hN8wYc4Lbe6Tj/Ou2
gP5sKi0nDqWx3Hr+M
zA9vCt7Bf1Ul6Ey4o
Js2Rn8Xa0Pd5Mk3Iq
4Wb7Lz1Tv9Co6Ge+H
X25519 key exchange AES-256-GCM keys never leave your devices nothing readable stored

Works with your stack

Bring the agent you already use.

antgrid doesn't replace your coding agent — it's the command centre over the ones you already run. No new model, no new CLI to learn.

Claude Code Codex opencode Cursor GitHub Copilot Kilo Kimi Mistral Vibe any terminal agent

How it works

1
Install on your machine
The desktop app runs your agents in real terminals and links them to your phone — Windows, macOS, Linux.
2
Run any agent
Start Claude Code, Codex or Cursor exactly how you do now. antgrid wraps them — no workflow change.
3
Take it with you
Watch, steer and approve from your phone. Step away — an AI supervisor keeps the run moving, checks the work against real evidence, and wakes you for the calls it can't make.

Get started

Let your agents run.
Check in when it matters.

Set up antgrid in a couple of minutes. Free on 2 machines, remote control included — you pay only to add machines and the AI supervisor.

Start free See pricing

No card required · remote control on the free plan

Download the desktop app

Download for macOS Download for Windows Download for Linux

App stores coming soon

App Store soon Google Play soon Mac App Store soon