Skip to main content

​
General

Whim is a cloud dev workspace for running AI coding agents. Each task runs in its own isolated container with a full development environment, terminal sharing, and git branch isolation. You write a prompt, pick an AI model, and watch it code in real time.
Whim handles all the infrastructure for you — cloud containers, git branch management, environment setup, and multi-agent orchestration. You can run multiple tasks in parallel without using your local machine resources, collaborate with teammates in real time, and switch between AI providers and models seamlessly.
Whim works in all modern browsers including Chrome, Firefox, Safari, and Edge. For the best experience, use a Chromium-based browser.

​
Tasks

It depends on your plan. Free and Pro plans support up to 20 active tasks per team. The Max plan supports 40+ active tasks, scaling with your capacity multiplier.
If a task container stops unexpectedly, Whim detects it automatically (typically within 30 seconds) and marks the task as stopped. You can review the terminal output to diagnose the issue, then fork or re-run the task with an updated prompt.
Yes. You can type in the live terminal to respond to AI agent prompts, run commands, or guide the agent. You can also send follow-up prompts to adjust the task’s direction.
A task launches immediately in its own container with an AI agent. A todo is planned work — it sits in your backlog until you’re ready to launch it. You can convert any todo into a running task.

​
Billing

CUs (compute units) combine container runtime and AI token usage. One CU equals roughly 30 minutes of container time or ~14,000 input tokens. See Compute Units for full details.
Yes. The Free plan gives you 50 CU per week at no cost. This is enough to run several tasks and explore the platform. Upgrade to Pro or Max when you need more capacity.
A grace period begins that lets running tasks finish briefly. You won’t be able to start new tasks until your weekly period resets or you upgrade your plan.
On the Max plan, you can increase your capacity multiplier (up to 500x) to scale your weekly CU allocation. You can also switch to annual billing to save 17%.

​
Security

Yes. Each task runs in its own isolated container with a separate filesystem and network stack. Your code is not accessible to other users. Workspace data is protected by row-level security at the database layer. See Data & Privacy for details.
Only members you’ve explicitly invited to your workspace can see its tasks, files, and settings. Whim enforces strict workspace-level access control.
Your code lives in your GitHub repository. Whim clones it into an ephemeral container for each task and destroys the container when the task ends. File attachments are stored securely in Supabase Storage, scoped to your workspace.

​
AI providers

Whim supports models from Anthropic (Claude), OpenAI (Codex), and a range of models via OpenRouter. You can select your provider and model when creating each task. See Supported Providers for the full list.
Whim provides built-in API access through its CU system, so you don’t need your own keys. AI costs are included in your CU consumption.
You select a model when you create a task. To try a different model, fork the task or create a new one with a different provider/model selection.

​
Collaboration

Go to Settings → Members and send an invite. Invited users will receive access to your workspace once they accept. See Invitations for details.
All workspace members can see and interact with every task in the workspace. You can assign tasks to specific members, leave comments, react, and use voice chat for real-time collaboration.
Yes. The live terminal supports multiple viewers. All workspace members can watch a task’s terminal output in real time and interact with it.