Plan mode: conversation context not visible to user + context-clear feature removed

Bug Report

Environment

• Claude Code v2.1.89
• Opus 4.6 (1M context) with max effort
• Linux (Ubuntu-based)

Issue 1: Conversation content not displayed to user

When using plan mode with multiple Agent (Explore) subagents for codebase exploration, the results from those agents are not visible to the user in the terminal UI. The user sees what appears to be compressed/truncated output, even though the context is NOT actually compressed.

Steps to reproduce:

1. Enter plan mode (EnterPlanMode)
2. Launch 3 parallel Explore agents for codebase analysis
3. Agents return detailed results (thousands of lines)
4. User cannot see the agent results in the terminal — they appear as if the conversation was compressed

The assistant has full access to the context internally, but the user's terminal shows nothing. This makes it impossible for the user to review the exploration results before plan approval.

Issue 2: Plan mode context-clear feature removed

Previously, plan mode had a feature to clear context and start fresh work after planning. This feature appears to have been removed in a recent update. The user relied on this to:

• Start implementation with a clean context after planning
• Avoid context window bloat from exploration phase

Impact

Both issues severely degrade the plan mode workflow. The user cannot see what the AI explored, and cannot start clean after planning. This makes the entire plan mode feature frustrating and nearly unusable for large refactoring tasks.

Expected Behavior

1. Agent/subagent results should be visible to the user in the terminal, or at minimum a clear summary should be auto-displayed
2. Context-clear feature in plan mode should be restored or an equivalent workflow provided