Cloud machine provisioning
Provide a token for a supported cloud, choose a region, size, image, and agents, and let Codex Remote create one server through a per-cloud OpenTofu module.
Codex Remote
Codex Remote provisions cloud machines, installs Codex and Claude Code, configures SSH, services, projects, and MCP servers, and manages the machines from a desktop app. It is intended for developers who want remote AI coding environments without repeating the setup manually.
Remote machines for Codex and Claude Code
Codex Remote is a desktop tool for putting Codex and Claude Code on remote cloud machines. It accepts a token from a cloud account you already use, provisions a server, installs the agents, configures access, and presents the resulting machine to the tools’ existing workflows.
The product packages the repetitive parts of remote-agent setup: SSH registration, non-interactive agent sign-in, boot-time services, project transfer, and MCP configuration. Machines from different providers can be managed from one list, while an MCP interface allows an AI agent to inspect or operate the environment when the relevant permissions are enabled.
Core features
Agent-ready server setup
The bootstrap process installs base packages, the Codex CLI at `/usr/local/bin/codex`, Claude Code under its own account, configured MCP servers, and a systemd unit for each agent.
SSH and agent registration
Each machine gets a named `Host codex-remote-<name>` entry in SSH config. Codex uses that host to start `codex app-server` over SSH, while Claude Code connects through its own account and login.
Machine status overview
The app tracks machines from different clouds in one list and shows whether each is paused, idle, or running agent sessions, helping indicate whether stopping a machine could interrupt work.
Project and MCP transfer
Projects are cloned or copied to the remote machine with untracked configuration sent separately, and portable MCP servers are carried over as part of the transfer.
Agent control and provider registry
An MCP server lets an external agent inspect machines and, when separately permitted, run commands, sync projects, create machines, or destroy them. Provider definitions can also be supplied through a custom registry.
Practical use cases
Starting a dedicated coding machine
Create a remote server with a selected cloud, region, image, and size, then have Codex and Claude Code installed and available without repeating the setup wizard by hand.
Managing multiple remote environments
Keep several machines across cloud providers in one list, with pause, idle, and active-agent status visible before stopping or resuming a server.
Moving a project to a remote host
Send a local project to a new machine while retaining the untracked configuration and portable MCP servers needed for the project’s tools.
Delegating infrastructure work to an agent
Connect an AI assistant through the Codex Remote MCP server so it can inspect machine health and, after permissions are enabled, provision, configure, sync, or remove infrastructure.
Using a custom provider catalogue
Point the app at a private provider registry to use a homelab module, an internal cloud definition, or a team-specific machine configuration without waiting for a desktop release.
Pros and Cons
Pros
- Automates server creation, agent installation, login setup, SSH registration, and boot-time service configuration.
- Supports project transfer with separate untracked configuration handling and carries portable MCP servers to the remote machine.
- Puts machines from multiple cloud providers into one status list and exposes estimated cost during machine creation.
- Uses separate permissions for reading, changing resources, and destroying machines when an AI agent manages infrastructure.
- Allows provider catalogues to be extended through a runtime JSON registry and custom OpenTofu modules.
Cons
- The available evidence does not establish pricing, plan limits, or a free tier.
- Creating and operating machines still uses the user’s cloud account and may incur provider charges.
- Destructive machine actions are supported but require explicit permission; destroying a machine cannot be undone.
FAQ
How do I install Codex Remote?
Install the desktop app with Homebrew using `brew install --cask pandelisz/tap/codex-remote`, or download the latest release, unzip it, and move it to Applications. The app updates itself after installation.
What happens when I create a machine?
Choose a supported cloud, enter its provider token, select a region, machine size, image, and agents, then create the machine. Codex Remote applies a per-cloud OpenTofu module, bootstraps the server over SSH, registers the host, and adds the required agent services.
Which cloud providers does it support?
The homepage lists Hetzner, AWS, DigitalOcean, Linode, Vultr, and Scaleway as cloud token options. The supplied provider registry contains a detailed Hetzner Cloud entry; provider availability and configuration can be refreshed through the registry.
Can it transfer my project and MCP servers?
Codex Remote carries over MCP servers and transfers a project together with its untracked configuration separately from the clone, so gitignored `.env` files are not silently omitted from the setup workflow.
Can an AI agent manage the machines?
Yes. An MCP server exposes tools for listing machines and sizes, running commands, syncing projects, and creating or destroying machines. Reading is available by default, while changing resources and destroying machines require separate permission switches.
Quick Facts
- Product type
- Remote development and AI agent machine management tool
- Primary agents
- OpenAI Codex and Claude Code
- Cloud options listed
- Hetzner, AWS, DigitalOcean, Linode, Vultr, and Scaleway
- Infrastructure workflow
- Per-cloud OpenTofu modules bootstrapped over SSH
- Desktop installation
- Homebrew cask or downloadable application
- Pricing
- Not established; the supplied pricing URL returns a 404
Alternatives à Codex Remote
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
GitBot
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.
Manta AI
Manta AI is an autonomous web app testing tool for teams that want to map application behavior, catch regressions, and generate tests without writing scripts or maintaining selectors. It works from a URL and supports plain-English test flows, run results with screenshots, and scheduled or deployment-triggered checks.
SonOf
SonOf connects to your repo and PM tool, audits the codebase and surrounding product context, and turns approved work into shipped tickets with senior engineering review. It is aimed at founders and engineering leaders who need backlog help without hiring a full team immediately.