Restore contextual input hint text (ghost text) that suggested next actions after Claude's response

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Several weeks ago, Claude Code's input box displayed a dimmed contextual hint (ghost text) after each response that suggested a relevant next action — e.g. "Try running the tests", "Ask Claude to explain this", or similar context-aware prompts based on what Claude had just done.

This hint was rendered with dimColor in the footer area, below the input prompt, and changed dynamically based on the session state.

The feature has silently disappeared. Searching the current cli.js source confirms that showSuggestedNextSteps — the settings key that previously controlled this — no longer exists in the compiled output. The setting is now a dead key that does nothing.

The input area currently only shows:

• ? for shortcuts (static, always shown when idle)
• hold [key] to speak (when voice is active)

The contextual hint that changed based on Claude's last response is gone entirely.

Proposed Solution

Restore the contextual ghost text in the input footer area that previously appeared after each Claude response.

Specifically:

1. Re-introduce the showSuggestedNextSteps setting (boolean, default true) so users can opt out
2. After each completed response, display a dimmed contextual hint in the input area suggesting a relevant follow-up action based on the session state
3. The hint should update dynamically — not be a static string like the current ? for shortcuts

The original implementation used dimColor text rendered in the same footer row as the shortcuts hint. That pattern still exists in the codebase — the ghost text logic just needs to be restored alongside it.

If the feature was intentionally removed due to performance or prompt caching concerns, please consider a lightweight alternative: a small rotating set of context-aware suggestions (e.g. "run tests", "explain this", "commit changes") gated by the showSuggestedNextSteps flag, rather than full AI-generated hints.

Alternative Solutions

No good workaround exists. The hint was a unique feature of Claude Code's TUI — no other tool or setting replicates it. The closest alternative is mentally tracking where you are in a task, which is exactly the cognitive load the hint was designed to reduce.

Priority

Medium - Would be very helpful

Feature Category

CLI commands and flags

Use Case Example

1. Working on a multi-file refactor across a large codebase
2. Claude finishes applying a set of edits
3. Previously, the hint would show something like "run your tests" or "review the diff" — surfacing the obvious next step without requiring the user to context-switch mentally
4. Now the input is blank and static — the user must remember where they were in their workflow

This is especially valuable during long agentic sessions where the user may have stepped away and returned mid-task.

Additional Context

_No response_