Skip to main content

Overview

A task is an AI agent session that runs in its own isolated container. You write a prompt describing what you want done, pick an AI provider and model, and Whim spins up a full dev environment where the agent works on your codebase.

Writing your prompt

The task composer sits at the bottom of the sidebar. Type or paste your instructions into the text area — it supports Markdown formatting, so you can use headings, lists, code blocks, and links to structure complex requests.
Good prompts are specific about the desired outcome. Instead of “fix the bug”, try “Fix the 404 error on the /api/users endpoint when the user ID contains a hyphen.”
Press Tab to toggle between creating a task (runs immediately) and a todo (saved to your backlog). Press Cmd/Ctrl + Enter to submit. Your draft is automatically saved per-workspace, so you won’t lose it if you navigate away.

Selecting an AI provider and model

Click the provider button (e.g. “Claude”) in the composer toolbar to choose between:
ProviderDescription
ClaudeAnthropic’s Claude, running natively via Claude Code
CodexOpenAI’s Codex, running natively via the Codex CLI
CCRCloud Code Runner — routes through OpenRouter for access to additional models
After selecting a provider, click the model button to pick a specific model. The list shows all models available for that provider, with your workspace default marked.
For Codex, each model option includes a reasoning effort level (Low, Medium, High, XHigh, or Default) that controls how much thinking the model does before responding.
Your workspace admin can set default provider and model preferences in Settings > Providers. These are pre-selected when you open the composer.

Adding file attachments

You can attach files to give the AI agent visual context, reference documents, or data files.
1

Attach files

Drag and drop files onto the composer, paste from your clipboard, or click the attachment button to browse.
2

Review attachments

Attached files appear as thumbnails below the prompt. Click the × on any attachment to remove it.
Limits:
  • Maximum 10 files per task
  • 25 MB per file
  • Supported types: images (PNG, JPG, GIF, WebP, SVG), PDFs, text files, code files (TS, JS, Python, Go, Rust, etc.), config files (JSON, YAML, TOML), and data files (CSV, TSV)
  • Executables (.exe, .dmg, .dll, etc.) are blocked

Choosing a source branch

Click the branch selector to pick which Git branch the task starts from. Whim fetches the list of branches from your connected repository.
  • The workspace default branch (usually main) is pre-selected
  • The task gets its own isolated branch created from the source branch you choose
  • If your workspace has tracked branch images, tasks launched from those branches use the pre-built environment for faster startup
The branch selector shows image status indicators when workspace images are available. An exact match means the image perfectly reflects the branch’s current configuration. A best-fit match means a compatible image is available but may be slightly out of date.

Task options

Mode

Toggle between Simple and Orchestrator mode:
  • Simple — the agent works directly on the task itself (default)
  • Orchestrator — the agent plans and delegates work to child tasks instead of implementing directly
See Orchestrator Mode for details.

Multi-try

Click the multi-try button to run the same prompt multiple times (up to 4 attempts). Each attempt runs as a separate task, letting you compare different approaches to the same problem. See Batches & Multi-Try for details.

Parent context

When creating a subtask under an existing task, toggle parent context to link the new task as a child. Child tasks inherit the parent’s branch context.

Launching the task

Click the Send button or press Cmd/Ctrl + Enter to launch. Whim will:
  1. Create an isolated Git branch from your selected source branch
  2. Spin up a container with your workspace’s dev environment
  3. Start the AI agent with your prompt and any attached files
  4. Stream progress to the terminal in real time
You cannot launch a task if your provider is not connected, you’re offline, or your compute unit balance is exhausted.

Monitoring progress

Once launched, the task appears in your workspace and its terminal streams live output. You’ll see:
  • Startup progress — container spawning, environment setup, and session initialization
  • Agent activity — the AI reading files, writing code, running commands
  • Status indicators — the task header shows the current agent state (Working, Idle, Needs Input, etc.)
You can interact with the terminal, send follow-up prompts, or pause the task at any time. See Task Lifecycle for details on managing running tasks.