Frequently Asked Questions

Foxpit is a desktop application (Windows, macOS, Linux) built with Electron and Three.js. It gives developers a visual cockpit to run and monitor multiple autonomous coding agents — such as Claude Code, Codex, and OpenCode — inside real terminals. Agents are visualized in a 3D isometric office, allowing you to see who is working, waiting for approval, or stuck at a single glance.

Not by the app. Foxpit's event hooks and terminal listeners connect to 127.0.0.1 on your machine, and the app collects no analytics. The app makes two kinds of outbound requests: signing in and checking your plan with the Foxpit account server (your email, an account id and an anonymous device hash, never code), and an update check against GitHub Releases. Keep in mind that the coding agents you run inside Foxpit still send your prompts and code to their own model providers — that is how those CLIs work.

Foxpit connects Claude Code, OpenAI Codex CLI, OpenCode and Antigravity in one click, and has its own built-in Foxpit engine that works with any provider you connect (Windows for now). Furthermore, because terminals are emulated via node-pty and xterm, you can run any standard CLI tool, bash script, or custom agent runner inside the cockpit.

No. The "Connect All" setup reuses Foxpit's bundled Electron/Node runtime, so it configures the necessary hook scripts without needing a global Node.js installation on your operating system.

When running multiple AI agents in parallel, they can easily overwrite each other's changes or create merge collisions in the main directory. Foxpit leverages native git worktree functionality: each sub-agent is assigned its own directory and isolated Git branch. Once an agent finishes its task, you can review the diff and merge cleanly.

Autonomous agents frequently pause to request confirmation before executing risky terminal commands or editing files. When this occurs, Foxpit immediately raises a visual banner in the 3D office, pulses the mascot desk, flashes the OS taskbar, and sends a system notification. Click the agent's status bubble or the summary bar to jump straight to that terminal.

Flagship models have strict token limits and higher pricing. Instead of letting the flagship model run slow compile commands or parse verbose test logs, Foxpit's orchestration workflow lets the lead agent assign sub-agents running cheaper models (such as a Haiku-class model, a free OpenCode model, or DeepSeek) inside optional worktrees. Because you pick each mascot's driver and model, you can keep expensive quota for planning and review.

Foxpit updates itself from GitHub Releases: a new version downloads in the background and installs when you restart. It also keeps your agent CLIs (Claude Code, Codex, OpenCode, Gemini, Antigravity) up to date; you can turn this off in Settings → Updates.

Yes, a free one. Sign in with an email code, Google or GitHub to start your 3-day free trial. No card needed. Your account only holds your email and plan; your code stays on your computer.

You will need Foxpit Pro to keep using the app: $5 for the first month, then $12 a month. Pro is not open yet: until it is, Foxpit keeps working after your trial and we will email you when Pro is available.

Yes. Every teammate runs on its own agent and model. Fox can plan on Claude Code while a teammate edits on Codex or a cheaper OpenCode model, all in the same project. Each one can work in its own Git worktree so they don't overwrite each other.

Tabs show text; Foxpit shows state. You see at a glance which agent is working, stuck or waiting for your approval, you get a notification when one needs you, and the lead agent hands work to the others instead of you copying prompts between tabs.

Yes. Teammates on the built-in Foxpit engine can use your own server (Ollama, LM Studio or any OpenAI-compatible address) or an API key from providers like OpenRouter, OpenAI, DeepSeek or Anthropic. The built-in engine runs on Windows for now; macOS and Linux are coming.