Preflight Checklist

• I have searched existing requests and this feature hasn't been requested yet
• This is a single feature request (not multiple features)

Problem Statement

When using specialized agents (launched via the Task tool), users cannot tell when Claude Code is operating in agent mode versus regular mode. This creates significant UX issues:

Core Problems:

1. Invisible Mode Transitions - Agents launch and complete silently with no visual indicator, leaving users confused about which system is responding
2. Lost Context Awareness - Users don't know if they're getting specialized agent capabilities or regular responses, leading to incorrect assumptions about answer quality and scope
3. Workflow Disruption - Tasks may be handled incorrectly when users assume an agent is still active but it has already completed and returned to regular mode
4. Manual Verification Required - Users must frequently ask "what mode am I in?" or infer from conversation context, breaking flow
5. No Session Context - Beyond basic directory and token warnings, users lack visibility into critical session state: active model, token usage progress, costs, active modes, etc.

Current State:

• Agent launches with no persistent UI feedback
• Agent completion happens silently
• No visual difference between agent vs regular responses
• Status line shows only: directory, token warning (>200K), session cost

Impact:

• Reduced trust in agent system
• Decreased productivity from context-switching confusion
• Missed opportunities to leverage specialized agents effectively
• Poor user experience compared to modern IDE status bars

Proposed Solution

Proposed Solution

Add a customizable status bar system with persistent agent mode indicator as the primary feature.

Core Features:

1. Agent Mode Indicator (Primary)

• Persistent visual indicator showing when agent mode is active
• Display format: 🤖 Agent: {agent-name} in status bar
• Nested agent support: 🤖 parent → child → grandchild
• Color-coded by agent type (using existing subagent color system)
• Mode transition notifications with configurable verbosity

2. Expanded Status Bar Widgets
   Beyond agent mode, create a modular widget system for comprehensive session context:

3. User Customization

Configure via slash commands

/statusbar preset minimal # Show only essentials
/statusbar preset verbose # Show everything
/statusbar add token-usage # Add specific widget
/statusbar remove cost # Remove widget

Or configure in settings.json

{
"statusBar": {
"widgets": ["agent-mode", "model", "token-usage", "cost"],
"format": "compact",
"notifications": {
"verbosity": "normal" # quiet | normal | verbose
}
}
}

4. Example Status Bar Outputs

Minimal preset

/project │ 🤖 general-purpose │ 45%

Default preset

/project │ 🤖 Agent: code-reviewer │ 90K/200K tokens │ $0.02

Verbose preset

~/dev/project (⎇ main) │ 🤖 Agent: general-purpose (2m 34s) │ sonnet-4.5 │ [████████░░] 80% │ ⏱ 1h 23m │ 💰 $0.0234 │ 💭 Thinking

5. Mode Transition Notifications
   🤖 Launching agent: general-purpose
   ⚙️ Agent mode active: general-purpose

[agent performs work]

✅ Agent general-purpose completed (2m 34s)
🔄 Returning to regular mode

Implementation Phases:

Phase 1 (MVP): Agent mode indicator + basic notificationsPhase 2: Full widget system + presetsPhase 3: Custom widgets via hooksPhase 4: API for programmatic access

Benefits:

• ✅ Solves core problem: always know which mode is active
• ✅ Extends to comprehensive session awareness (tokens, cost, model, etc.)
• ✅ User-configurable for different workflows
• ✅ Follows existing patterns (hooks, status line, subagent colors)
• ✅ Non-breaking: defaults preserve current behavior

FEATURE_PROPOSAL_CUSTOMIZABLE_STATUS_BAR.md

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