Conversation ​

Conversation Methods ​

After creating a task, you'll continuously interact with the agent in the central area. The usage here is basically consistent with the interaction habits in the IDE sidebar - centered around "entering requirements, adding context, viewing execution process, and continuing to ask questions."

Common operations include:

  • Enter requirements or additional explanations directly in the input field.
  • Click the send button or press Enter to send messages.
  • Continue asking questions in existing tasks, letting the agent continue processing with previous context.
  • View replies, results, and intermediate steps during task execution.

Task conversation interface example

Top Bar Actions ​

Besides displaying the current task title, the top of the conversation area provides some commonly used actions for quick switching and maintaining context during task processing.

  • Open Editor: Quickly open the editor or project content associated with the current task, allowing you to view the conversation while returning to code or files for continued processing.
  • Keep Workspace: When continuously asking questions or making repeated modifications in the same project, you can keep the current workspace to reduce repeated switching and reselection.

Sending Messages and Follow-up Questions ​

In the input field at the bottom of the conversation area, you can send new messages at any time.

Content suitable for direct sending includes:

  • New requirements: e.g., adding a feature or supplementary explanation.
  • Modification suggestions: e.g., adjusting implementation approach, adding constraints.
  • Follow-up questions: e.g., asking the agent to continue optimizing based on current results.

If you're already accustomed to the input, send, and follow-up workflow in the IDE sidebar, the experience here is basically the same - no relearning required.

File and Image Upload ​

Besides direct text input, you can also add files or images as context:

  • Paste Screenshots: Use Ctrl/Cmd + V to paste images directly from clipboard.
  • Upload Files: Click the upload button, or drag files directly into the input field area.
  • Add Local Content: Combine files, images, and text descriptions to help the agent understand the problem faster.

Execution Process Display ​

When the agent is executing a task, it displays key steps and results in the conversation area so you can track progress at any time.

You'll typically see:

  • Agent's reply content
  • Stage descriptions during execution
  • Generated result summaries
  • Expandable intermediate steps

The benefit is that you don't need to leave the current task - you can continuously view input, process, and output in the same place.

Interruption and Continuation ​

If a task is executing, the input field area will show a stop button, allowing you to interrupt the current execution at any time.

After interruption, you can still:

  • Continue adding new explanations
  • Adjust requirements and resend
  • Continue progressing the task based on existing content

This is consistent with the common sidebar-style interaction in the IDE - the focus is on maintaining continuous conversation rather than starting over each time.