Run Goose on Your Phone (iPhone + Android)

You can run goose on your phone today, iPhone or Android, but not by running the agent on the phone. goose is an open-source command-line AI agent that needs a real Linux shell. The fastest path: install Cosyra for iOS or Cosyra for Android, open the Ubuntu container, install the goose CLI with curl -fsSL .../download_cli.sh | bash, run goose configure with your own provider key, then goose session. goose is not one of the agents we pre-install, so this is the guide where you add it yourself. 1 hour free on Cosyra signup, no credit card.

This guide was written by the Cosyra team. We installed and ran goose inside a Cosyra container from a phone, and we cross-checked every claim about goose against the official aaif-goose/goose repo, the goose installation docs, and the Goose AI App Store listing, verified 2026-06-08.

What is goose?

goose is an open-source, extensible AI agent that runs on your machine and goes beyond code suggestions: it can install, execute, edit, and test with any LLM you bring. It started at Block, then moved to the Agentic AI Foundation under the Linux Foundation, so the repo now lives at aaif-goose/goose — about 47,600 GitHub stars, Apache-2.0, latest release v1.37.0 (2026-06-03) as of 2026-06-08. It ships a desktop app and a CLI for macOS, Linux, and Windows, and it is extensible over MCP.

The thing that defines goose, and the reason it is worth running from a phone at all, is that it is an autonomous agent rather than an autocomplete. You describe a goal, and it plans the steps, edits files, runs shell commands, and checks its own work. That loop wants a real shell with a package manager and network access. A phone keyboard is fine for kicking off that loop; what you actually need is somewhere for the loop to run.

The opinion we hold that the self-hosting crowd will push back on: the honest answer to "I want goose on my phone" is to stop running the agent on hardware you have to babysit. The "keep a beefy desktop on 24/7 and tunnel into it" setup is a real, free option, and for someone who already runs that machine it is genuinely nice. But we would rather not own, power, patch, and expose a home server just to check on an agent from the train. Put the agent on a cloud box that is already always on, and the phone becomes the only device you carry.

How can you run goose from a phone?

You can reach goose from a phone three ways, and the deciding factor is where the agent process lives: a cloud Ubuntu container reached from a native app (Cosyra), the Goose AI iOS app pointed at your own always-on machine, or an SSH client into a VPS you run. Comparison current as of 2026-06-08.

1. Cosyra: run the goose CLI in a cloud container

This is what we built: a native iOS and Android terminal connected to a persistent Ubuntu 24.04 container. Claude Code, Codex CLI, OpenCode, and Gemini CLI are already installed; goose is the one you add with the official installer. The container is always on, so the agent has somewhere to run and keep running while your phone sleeps.

• Works when: you want goose running in the cloud, reachable from both iOS and Android, with no home server to keep powered and patched.
• Breaks when: you are offline. The container lives in the cloud, so no internet means no terminal. Full trade-off list in Cosyra vs Termux.
• Cost: 1 hour free on signup, no credit card. Extend with a 10-hour, 7-day trial when you want more. After that, $29.99/month or $300/year. Your model tokens are billed by your provider. See pricing.

2. Goose AI iOS app pointed at your own machine

The free Goose AI app is a remote control. You run the goose agent on a desktop or server you own, expose it through a tunnel, and the app connects back over HTTPS to watch and drive it. The goose-mobile README describes the iOS client as "an implementation of a remote protocol to access the goose agent" that "connects back to your goose agent from anywhere, via a tunnel."

• Works when: you already keep a powerful machine on around the clock and want a free way to check on long-running jobs from your phone.
• Breaks when: that machine sleeps, reboots, or loses its tunnel. The phone is only a window; the agent depends entirely on your host staying up.
• Cost: free app, free goose, plus the electricity and upkeep of an always-on machine and a tunnel you maintain.

3. SSH from an iOS client into your own VPS

On iPhone or iPad, an SSH client into a VPS works. Spin up a box, install the goose CLI on it, SSH in inside a tmux session, and run goose session. Most VPS images are x86_64, so the installer pulls the same prebuilt binary it does on Cosyra; the difference is you are now the sysadmin.

• Works when: you want a box that is yours and an iOS-native keyboard.
• Breaks when: you do not want to patch, harden, and back up a server. More in SSH from your phone.
• Cost: the SSH client plus the VPS (about $5 to $40/month). No Android equivalent for the iOS-only clients.

How do you set up goose on iPhone or Android?

You set up goose on iPhone or Android in about five minutes with Cosyra: install the app, open the container, install the goose CLI, configure your provider key, and run a session. The install is one command because the container is x86_64 and the installer ships a prebuilt binary.

Step 1: Install Cosyra and open a container

Download from the App Store or Google Play. Sign in with Apple, Google, or email. On first launch we provision a fresh Ubuntu 24.04 container with Node.js, Python, Git, and tmux already on it. This is the always-on machine goose runs on, so there is nothing to keep powered at home.

Step 2: Install the goose CLI

goose is not one of the four agents we pre-install, so you add it with the official installer. On x86_64 this pulls a prebuilt binary, so it is a download, not a build. We ran this in a fresh container and goose --version reported 1.37.0.

Step 3: Configure your LLM provider

goose is BYOK and model-agnostic. Run goose configure, choose a provider, and paste your own key. The config lands in the container's home volume, so it survives hibernation and is there on every device you sign in from.

Step 4: Start a goose session

Clone your repository, cd into it, and start a session. Describe the change; goose plans, edits, runs commands, and tests, then reports back. Because the container is persistent, you can close the app mid-task and reopen later to pick up where it left off.

What can you actually do with goose on your phone?

The honest pitch for goose on a phone is an autonomous agent that keeps working while you are not looking at it. Three sessions we run from a phone.

Kick off a task on the train and read the result later

You have a self-contained chore: bump a dependency and fix what breaks. On the train, open Cosyra, cd into the repo, run goose session, and describe it. goose works through the change while you put the phone away. Because the container stays up, the diff and test output are waiting when you check back at the next stop.

Triage a failing test from the couch

Saturday morning, phone in hand, CI is red. Point goose at the failure: "the test in tests/test_invoice.py is failing, find out why and fix it." It reads the traceback, edits the code, reruns the test, and tells you what it changed. You review the diff on the phone and decide whether to keep it.

Keep a private repo inside the container

Because goose is BYOK and the container is yours, a sensitive repo can stay inside the box. Point goose at the provider you trust, and the code never leaves the container for anything except the model calls you configured. For the cross-agent view of how this compares to the agents we pre-install, see the AI coding agents on mobile pillar.

What are the real limits of running goose on a phone?

Knowing where this setup stops helps you match it to the right job instead of fighting it.

• No offline mode on Cosyra. The container lives in the cloud, so no internet means no goose. If you code on planes with no wifi, an agent on local hardware is the only thing that runs there.
• Not pre-installed on Cosyra. Unlike Claude Code, Codex CLI, OpenCode, and Gemini CLI, you install goose yourself. It is one command, but it is a command. See OpenCode on phone for the other open-source agent, which we do pre-install.
• The iOS remote-control app is a different thing. If you install Goose AI from the App Store expecting an on-phone agent, you will be confused. It needs a machine you host. Running the CLI in a container is the setup this guide covers instead. Jules carries the same trap: its only mobile client is an unofficial third-party app that browses sessions, not a shell into the agent.
• You pay your provider. goose is free, but the tokens are not. Budget for API spend the same way you would on a laptop.
• No Docker-in-Docker guarantee. If a task goose runs needs docker, verify it works before committing to phone-only.

Who should pick each option?

Choose the Goose AI iOS app if you already keep a powerful machine on around the clock, you want a fully self-hosted agent, and you do not mind running a tunnel. Choose an SSH client plus a VPS if you are on iOS, you already run a box, and you want full control of the host. Choose Cosyra if you want goose running in the cloud on both iOS and Android without maintaining a machine, and you want Claude Code, Codex CLI, OpenCode, and Gemini CLI sitting next to goose in the same container. We weigh that trade-off in full — a self-hosted local agent with an iOS remote against a hosted cloud container with native apps — in Cosyra vs goose, including what a "free" self-hosted stack actually costs once you price the always-on host.

For the same walkthrough with other user-installed CLIs, see Aider on phone and Qwen Code on phone.

Frequently asked questions