Add context-aware suggested next prompts to the composer

What variant of Codex are you using?

Codex CLI 0.153.0. The request could also benefit the App and IDE Extension composers.

What feature would you like to see?

Add an optional context-aware suggestion for the logical next user prompt after Codex completes a turn.

For example, after implementing a change, Codex could suggest:

Run the full test suite and fix any failures.

The suggestion could appear as ghost text or a suggestion chip in the composer. A user should be able to accept it with Tab or Right Arrow, dismiss it with Escape, or simply start typing something else. It must never run automatically.

This is similar to the next-step suggestions available in Claude Code and would reduce repetitive typing during common agentic loops such as implement, test, review, and commit.

Ideally this would be opt-in through a setting and generated from the current conversation and workspace state, so it recommends a concrete next action rather than a generic canned prompt.

Additional information

There currently appears to be shell completion and support for reusable custom prompts, but no native option for dynamically predicting the next conversational prompt. A static AGENTS.md instruction can make Codex print a recommendation in its response, but it cannot prefill or offer that recommendation directly in the composer.

Start by locating the Codex CLI composer and reviewing the existing shell-completion and reusable custom-prompt flows. Define how an opt-in suggestion is generated from conversation and workspace state, how it is accepted or dismissed, and how to verify that it never runs automatically.