Include dismissed prompt suggestion as context for the model

Problem

When Claude Code suggests a follow-up prompt and the user dismisses it by typing their own text, the model has no awareness of what was suggested. This creates confusion when the user's response is a reaction to the suggestion (e.g., questioning or correcting it).

Example

1. Claude Code suggests: "there is one at the root of the project"
2. User dismisses it and types: "is there?"
3. The model has no idea what the user is referring to

Proposed Solution

Include the dismissed prompt suggestion as context in the conversation so the model can understand what the user is reacting to. For example, a system note like: "Suggested prompt (dismissed by user): ..."

This would improve conversational coherence without requiring any change in user workflow.