Feature Request: Built-in Output Paging for Interactive Mode

Feature Request: Built-in Output Paging for Interactive Mode

Problem

Claude Code currently lacks built-in paging capabilities for long outputs in interactive mode. Users must rely entirely on their terminal emulator's scrollback buffer to review:

• Long file contents
• Multi-file search results
• Detailed code analysis
• Comprehensive documentation
• Extended conversation history

This creates a poor user experience, especially for common software engineering tasks that frequently produce lengthy outputs.

Current Workarounds (All Suboptimal)

1. Terminal scrollback navigation - Requires switching mental context from the conversation to terminal navigation (Shift+PageUp, etc.)
2. Manually requesting file dumps - Users must explicitly ask Claude to save outputs to files, then exit to view them
3. Truncated information - Users miss important details that scroll off-screen

Expected Behavior

Claude Code should provide one or more of the following:

Option 1: Automatic Paging for Long Outputs

When output exceeds a configurable threshold (e.g., terminal height), automatically pipe through a pager:

• Default to less -R (preserves ANSI colors)
• Respect $PAGER environment variable
• Allow configuration in settings.json

Option 2: Interactive Pagination

Implement built-in pagination similar to git log:

• Display first page of output
• Show prompt: -- More -- (space to continue, q to quit, / to search)
• Standard navigation: space/pagedown, b/pageup, j/k for line-by-line

Option 3: Configurable Auto-File Dumping

Add setting to automatically write long outputs to temporary files with notification:

Output is long (500 lines). Saved to /tmp/claude-output-12345.txt
View with: less /tmp/claude-output-12345.txt

Suggested Configuration

{
  "output": {
    "paging": {
      "enabled": true,
      "threshold_lines": 50,
      "pager_command": "less -R",
      "auto_dump_to_file": false
    }
  }
}

Use Cases

1. Reading long source files with Read tool
2. Reviewing multi-file Grep search results
3. Examining detailed agent task outputs
4. Reading comprehensive documentation responses
5. Reviewing conversation history

Impact

This affects all interactive Claude Code users who work with:

• Large codebases
• Detailed analysis tasks
• Documentation exploration
• Multi-file operations

Related Tools

Most interactive CLI tools handle this well:

• git log - built-in pagination
• man - uses less by default
• docker logs - supports --follow with paging
• kubectl logs - similar paging support
• Python REPL help() - auto-pages long help text

Priority

High - This is a fundamental usability issue for an interactive developer tool. The current experience forces users to interrupt their workflow to manage output visibility, which degrades the conversational nature of the interface.

Environment

• Claude Code version: latest as of 2025-11-23
• OS: macOS (affects all platforms)
• Terminal: Multiple (iTerm2, Terminal.app, etc.)