AgentsDock

AgentsDock is the control surface for the AI agents you already run.

AgentsDock talks to a small open-source server (AgentsServer) that you deploy yourself in one command. Every chat, job, and file stays on your infrastructure — and can be reached from anywhere.

SYNCED ACROSS EVERY DEVICE
Start a job on your Mac, check it from your iPhone or iPad, or drive it from Linux — every chat stays in sync in real time, wherever you are.

RICH MEDIA, RIGHT IN THE CHAT
Videos, simulation output, and image grids land directly in the conversation as your agent produces them.

ALL YOUR CHATS, ORGANIZED
Group work by project or folder, switch between Claude and Codex per chat, and keep every thread in one place instead of scattered across terminal tabs.

FULLY SELF-HOSTED
Run everything on your own Mac, Linux box, or cluster. Your code, credentials, and outputs never leave infrastructure you control.

FULL TERMINAL, EVEN ON YOUR PHONE
Each chat owns a persistent terminal session on your server — drop into a real shell from your phone whenever you need to.

SCHEDULE JOBS AHEAD OF TIME
Queue a run to fire now, after an interval, or on a repeating loop — kick off long research jobs without babysitting them.

HAND OFF BETWEEN AGENTS
Digest mode lets one chat summarize its work and send it straight into another, so long chains of work don't need manual copy-pasting.

OPEN SOURCE SERVER
AgentsServer is MIT-licensed and fully open source — read the code, fork it, and self-host it for as long as you like. The client is not open source yet.

Requires a server you control running the AgentsServer installer, with Claude Code and/or Codex CLI already authenticated. AgentsDock does not include AI models or API access itself.