[FEATURE] Option to manually edit Claude's Plan

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 Claude generates a plan, users cannot directly edit these outputs to refine them. Instead, users must:

• Read through Claude's entire response
• Identify what needs to be changed
• Write a new prompt explaining the desired modifications
• Wait for Claude to regenerate the entire response
• Consume additional tokens for the regeneration

This creates several issues:

• Token waste: Even minor edits (changing a filename, adjusting a single step, reordering tasks) require full regeneration
• Time inefficiency: Multiple back-and-forth exchanges to get a plan exactly right
• Context bloat: Each iteration adds to the conversation history, consuming more context window
• Friction: Interrupts the creative flow when users want to quickly tweak a plan before execution

Proposed Solution

Add an inline editing capability for Claude's Plan Mode responses.

User Interface

When Claude presents a plan, provide an option to Edit Plan:

Claude's Plan:
1. Set up the database schema
2. Create the API endpoints  
3. Implement authentication
4. Build the frontend

[Edit Plan] [Yes and auto-accept edits] [Yes and manually approve edits] [No, keep planning]

Clicking [Edit Plan] would:

• Open the plan in an editable text buffer (similar to how Ctrl+G works for user prompts)
• Allow users to modify text directly
• Save changes back to the conversation context
• Treat the edited version as Claude's response when plan is accepted

Alternative Solutions

Option A (Minimal): Edit in external editor

• Similar to existing Ctrl+G for user prompts
• Opens Claude's last response in $EDITOR
• User makes changes and saves
• Modified version replaces original in context

Option B (Enhanced): Inline TUI editor

• Use terminal UI to highlight and edit specific sections
• Arrow keys to navigate, Enter to edit, Esc to save
• Visual indicators for what's been modified

Option C (Hybrid): Edit-on-demand command

• Add /edit-response command
• Specify line numbers or section: /edit-response lines 3-5
• Opens just that portion for editing

Priority

Medium - Would be very helpful

Feature Category

Interactive mode (TUI)

Use Case Example

1. Reordering tasks: User wants step 5 done before step 3
2. Filename adjustments: Claude suggests helper_utils.py, user wants utils/helpers.py
3. Scope changes: Removing one feature from a multi-feature plan
4. Priority adjustments: Marking certain tasks as "optional" or "phase 2"
5. Combining steps: Merging two related tasks into one

Additional Context

Technical considerations

• Edited responses should be clearly marked in conversation history (e.g., "[User-edited]")

Priority Justification

• Addresses a common pain point (iterating on plans)
• Significant token savings potential
• Aligns with Claude Code's goal of efficient, collaborative development
• Relatively contained scope (doesn't require model changes)

This feature would make Claude Code feel more like pair programming, where you and Claude can both edit the "shared document" (the plan) before executing it, rather than having to verbally describe every change.

In the future this could be similarly expanded to allow edits to any text-block response, allowing the user to keep context focussed and correct, reducing token usage, and preventing context degradation.