[FEATURE] Add a Prompt Bench for Saving and Restoring Drafted Prompts

Resolved 💬 2 comments Opened Apr 28, 2026 by 69Jesse Closed Apr 28, 2026

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

When using Claude Code in a long-running development session, users often think of follow-up feature requests while Claude is still implementing the current feature. These follow-up prompts may depend on the current feature being completed correctly.

Today, the user has to either keep the new prompt somewhere outside Claude Code, risk submitting it too early, or leave it sitting in the prompt bar while continuing to refine the current implementation. This creates friction because the user may not want Claude to see, interpret, or act on the new request until the current feature is working as intended.

This is especially problematic when the next prompt assumes a stable state. If the current feature still has bugs or needs refinement, sending the next feature request too early can confuse the implementation flow, compound errors, or force the user to restate context later.

Proposed Solution

Add a “prompt bench” feature to Claude Code that lets users temporarily save one or more drafted prompts without submitting them.

A user should be able to write a prompt, bench it, continue working on the current feature, and later restore the benched prompt into the prompt input when they are ready to send it. The bench should support zero or more saved prompts, and users should be able to add prompts to the bench, view benched prompts, remove prompts from the bench, and restore any benched prompt back into the prompt bar at any time.

This would let users prepare future instructions while Claude is working, without disrupting the current implementation flow. Once the active feature is correct, the user can pull the next prompt from the bench, optionally edit it, and submit it with the current project state as context.

The expected behavior is:

  1. Draft a prompt in the input bar.
  2. Move it to the prompt bench without sending it to Claude.
  3. Continue refining the current task.
  4. Open the bench later and select a saved prompt.
  5. Have that prompt automatically filled back into the input bar.
  6. Submit, edit, reorder, or delete benched prompts as needed.

This would improve multi-step development workflows by separating “prompts I am preparing” from “instructions I am ready to give Claude now.”

Alternative Solutions

_No response_

Priority

High - Significant impact on productivity

Feature Category

CLI commands and flags

Use Case Example

_No response_

Additional Context

_No response_

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗