Frequently asked questions

What coding agents can I run?

Whatever's currently #1 on SWE-bench. Today that includes Claude Code, Codex, Gemini CLI, and Cursor CLI. We add more as they ship.

What codebases can I connect to?

Anything. Connect any existing git repo or start fresh from a template. We auto-set-up and run your project regardless of stack.

How much usage do I get?

Unlimited agent messages on paid plans. Connect your own provider account (Claude, ChatGPT, Gemini, Cursor) and never worry about token costs while coding.

Where do agents actually run?

On secure remote computers we provision via Freestyle. Each computer is its own sandbox with a stable preview URL.

Where can I prompt agents from?

iMessage, Slack, Discord, WhatsApp, and the web canvas. We're always adding more.

Can I view the code the agent writes?

Yes — preview diffs, logs, and the rendered preview side by side. You can also test the running app from any device.

Who gets credit for the commits?

You do. Configure exactly which GitHub account gets attribution under Project settings → Git.

How are you different from Codex Cloud or Claude Code Cloud?

None of those serve previews, so you can't actually test what the agent built. We run real previews, support all the top agents in one place, and have the deepest iMessage integration.

Are you open source?

Soon. We plan to open source the templates first, then the agent runner over time.

What's next?

Closing the loop on agent-driven testing (test runs, recorded videos), deeper deploy integrations, and skills that capture the latest tricks the top coding agents are using. The future will be built on grasscoding.