Buying
Launch and license
When can I buy it? ›
Soon — sales open after the launch polish pass. The waitlist gets one launch email.
What will it cost? ›
Pricing and licensing are being finalized and will be announced at launch. The waitlist hears first.
How do I report a bug or request a feature? ›
Raise a ticket in the app or on the web and the developer works on it. The ticket system is built and deploying now.
Setup & providers
Install, models, servers
What OS does it run on? ›
Windows first. The core is portable, but Windows is the launch target.
Is this the same as Claude Code? ›
No. A single CLI is one worker; CrazyToad is the coordinating brain that runs many workers in parallel.
Do I need separate API keys? ›
Not for CLI-based Claude, Codex, and Gemini — your existing subscriptions carry the work. API keys are optional lanes for OpenAI-compatible providers, xAI, and remote routing.
Can it use local models? ›
Yes. Local Ollama lanes live in Accounts and can be routed for jobs that fit.
Can I run agents on my own servers? ›
Yes. A one-line install turns your own Linux box into a hardened agent — confined to a workspace, non-root, behind a deny-list and a daily cost cap. Run it on as many of your own hosts as you like.
Privacy
Local-first posture
Does CrazyToad send my code anywhere? ›
No — zero telemetry. Provider calls go through the accounts and CLIs you configure, the same trust boundary as using those providers yourself. App state stays local unless you deliberately send data through a configured provider or delete that local state.