[Feature Request] Persistent shell output access to avoid redundant long-running command executions

Problem Statement

When working with long-running tests (tens of seconds to minutes), Claude Code's shell output handling is inefficient:

1. Output is truncated or not fully accessible after execution. Ctrl+O expansion doesn't help for big output, it also truncates too much.
2. Claude runs a test, greps for "Error", then re-runs the same test to grep for different keywords
3. All relevant debug output often existed in the first run but wasn't captured

This wastes significant time when debugging integration tests or SBF tests.

Proposed Solution

Automatically persist shell command output to a file that Claude can grep/read multiple times without re-executing.

Alternative approaches:

• Allow Claude Code to connect to a separate terminal for full output visibility
• Cache recent command outputs so Claude can search them without re-running

Alternative Solutions

Current workarounds:

• Run tests manually in a separate terminal and copy output to Claude
• Pipe shell output to a file and provide Claude the file path

Priority

High

Feature Category

Terminal / Shell