What it does

AgentsMesh is a control layer for running a fleet of terminal-based AI coding agents. You install a self-hosted runner on each machine, and every agent gets its own pod holding a separate Git worktree, private credentials and its own branch, so concurrent agents do not interfere with each other. Pods can be bound together and talk over channels with @mentions. An Autopilot mode lets a control agent watch a pod and send the next instruction when it goes idle, with iteration limits and human takeover. Tickets sit on a Kanban board and can be tied to a pod, with merge or pull request tracking. Built-in agents are Claude Code, Codex CLI, Gemini CLI, Aider and OpenCode, and any other terminal-based agent can be used. The console is available on the web, as a desktop app and on iOS. You can use the hosted service at agentsmesh.ai with your own AI API keys, or run the stack yourself from Docker images. The code is under the Business Source License 1.1; the README says production use needs a commercial license until the 2030-02-28 change date, after which it becomes GPL-2.0-or-later.

Capabilities

Self-hosted runners that schedule agents across a fleet of machines
Per-agent pods with an isolated Git worktree and private credentials
Autopilot control agent that sends the next instruction when a pod goes idle
Channels and pod bindings so agents can collaborate with @mentions
Kanban tickets linked to pods with merge or pull request tracking
Web, desktop and iOS consoles with live terminal streaming

When to use it

You want to run many coding agents in parallel without them colliding
You need agents to run on your own infrastructure while you steer them from one console
You want a ticket board connected to agent sessions