Prompt suggestions appear as if they are part of the assistant's response

Description

The auto-generated prompt suggestions that appear below the assistant's response are not part of the context.

Example

After the assistant responds, suggestions like "enable automatic deploys on merge to main" appear below the response. This was actually a dumb idea for my project, so I copied it and asked the model why it suggested it.

It had no idea what I was talking about.

Expected behavior

The next course of action should be part of the context so the model can answer questions about it's thinking about the next steps when asked.

Environment

• Claude Code CLI
• macOS