Skip to main content
The more context you can provide Clark on where / what to build, the better. For example, to iterate on a specific feature after an initial generation, you can target or tag the relevant components so Clark knows exactly what parts of the application to focus on (e.g., sidebar, specific div, backend API). Similarly, in addition to prompting with text alone, uploading files (e.g. screenshots, CSVs) is an effective way to provide Clark with more information on the requirements at any point in a chat.

​
Targeting frontend elements

To focus Clark on any specific parts of the UI, select the Target icon in the chat or use keyboard shortcuts (hold ⌘ on Mac or Ctrl on PC).
Enter target mode with Clark
After entering target mode, select any element on the canvas to add it to the chat context before submitting a new prompt. Targeted elements are highlighted in purple and added to the chat context. Selecting components in target mode

​
Tagging backend APIs

Similar to integrations, you can focus Clark on specific APIs by @mentioning them in chat. Tag API in chat with @

​
Uploading files

To add more detail to your prompts, upload files like images, JSON, CSVs, and more. For example, include a screenshot or wireframe to convey layout, hierarchy, and component states. You can attach files from the Clark chat in the editor and from the Clark prompt on the homepage.
Attach files in chat

​
File size limits

  • Images: up to 50 MB raw. Large images are automatically resized so the longest edge is at most 1568 pixels before Clark receives them.
  • Text files (for example JSON, CSV, Markdown): up to 1 MB.