/task — Autonomous Task Execution

The /task command tells StarZero Agents to work on a complex job independently, from start to finish. Instead of a back-and-forth conversation, you describe what you want and the AI takes it from there.

How to Use It

Type /task followed by a description of what you want:

/task Create a 60-second highlight reel about product launches from my library

/task Find the 5 best product demo moments, create a project with them, add word-by-word captions, and render in 9:16

/task Create 3 non-overlapping podcast cutdowns from this episode, each about 45 seconds

What Happens Next

Once you send a /task command:

1. The Task Agent takes over. A specialized agent designed for autonomous, multi-step work picks up your request.
2. A to-do list appears. The agent breaks your request into steps and displays them as a tracked to-do list in the chat.
3. The agent works through each step. You can watch as each task moves from pending to in-progress to complete.
4. The result lands on your timeline. When the task is done, your project is ready to review.

Watching Progress

As the task runs, you'll see the to-do list updating in real time. Each step shows its status, so you always know where things stand. You can expand the agent activity sections to see exactly what's happening at each step — what searches were performed, which clips were selected, and why.

Suggested Actions

When a reply finishes, a Suggested actions row can appear at the bottom of the chat with one-click next steps based on what the reply produced:

• Proceed with plan — Shown after the AI outlines a plan and stops for your go-ahead. Clicking it tells the AI to execute the plan it just described, so you don't have to type out a confirmation — review the proposed steps and click when you're ready. The AI keeps replying in the language you've been chatting in, even though the button itself is labeled in English.
• Open editor — Shown when the reply changed or rendered your video project and the editor isn't already open, so you can jump straight to the timeline and see the result.
• Show files — Shown when the reply created files, opening the Artifacts panel so you can preview or download them.

The row only offers what's relevant, so you'll often see one button, sometimes none.

Stopping a Task

Changed your mind? Need to go a different direction? Click the stop button in the chat to halt the running task. The AI will stop after completing its current step. Anything already added to the timeline stays there — you can undo it or build on it.

Messages During a Task

A task runs hands-off, so it doesn't take new messages along the way. In a regular chat you can send messages while StarZero Agents is working; during a task, a message you send is turned away with a notice asking you to stop the task first, and your text stays in the message box. Stop the task if you need to change direction, then send your message.

A task also needs an idle chat to start: sending /task while a reply is still in progress shows a notice, and the command stays in the message box until the reply is done.

What Tasks Can Do

Tasks can handle multi-step workflows that combine several capabilities:

• Search your library for specific content
• Analyze clips for quality, relevance, and storytelling potential
• Build video projects with selected clips on the timeline
• Add captions in your preferred style
• Render the final output in your chosen aspect ratio
• Quality check the rendered result and make fixes

All of this can happen in a single task, without you needing to intervene at each step.

When to Use /task vs. Regular Chat

Use /task when...	Use regular chat when...
You have a clear, complete request	You want to explore and iterate
You trust the AI to make decisions	You want to approve each step
The job has multiple steps	You have a quick question or small edit
You want hands-off production	You want a collaborative conversation

Tips for Better Tasks

• Be specific and complete. Include duration, aspect ratio, topic, and any other requirements in your task description. The more detail you give upfront, the better the result.
• Include quality criteria. "Find the 5 most energetic moments" is better than "find 5 moments."
• Specify the output format. "Render in 9:16 with word-by-word captions" removes ambiguity.
• Start with simpler tasks. Get a feel for how the AI interprets your instructions before sending complex multi-part tasks.
• Review before sharing. Even though the task runs autonomously, always review the result on the timeline before distributing the final render.

If Something Goes Wrong

Tasks don't always produce exactly what you envisioned — and that's fine. After a task completes:

• Review the result on the timeline
• Ask for changes in the chat: "The second clip is too long — trim it down"
• Undo everything if you want to start over
• Try a different approach with a more specific or differently worded task

The beauty of working with AI is that iteration is fast. Don't be afraid to run another task with adjusted instructions.