Provide visibility into skills used by subagents, especially Plan subagent

Preflight Checklist

• [x] I have searched existing feature requests and this is a new, single feature request
• [x] This is not a bug report or support request

Problem Statement

When subagents (particularly the Plan subagent) execute, there's no clear visibility into which skills they're using. This creates several challenges:

• Lack of transparency: Users can't see what the subagent is actually doing under the hood
• Difficult debugging: When things go wrong, it's hard to diagnose which skill failed or behaved unexpectedly
• Missed learning opportunities: Users can't discover which skills are available or how they're being used
• Opaque decision-making: The subagent's thought process and skill selection logic remains hidden

This is especially problematic with the Plan subagent, which may chain multiple skills together without any indication to the user.

Proposed Solution

Display the skills being invoked by subagents in the output, similar to how tool calls are shown for the main agent.

For the Plan subagent specifically, show:

• Which skills are being considered for the task
• Which skills are actually executed
• Skill execution results or brief summaries
• Any skill failures or fallbacks

This could be presented as:

• A compact inline notation (e.g., [Plan Agent → typescript skill])
• An expandable section showing skill invocations
• Integration with the existing tool call display mechanism

Priority

Medium

Feature Category

Other

Additional Context

This feature request is related to issue #10122, which addresses visual differentiation of skills/agents/commands using colors and icons.

Key difference:

• Issue #10122 focuses on HOW to present information visually (colors, icons, formatting)
• This issue focuses on WHAT information should be shown (which skills are running)

These are complementary improvements that solve different aspects of the same visibility problem. Together, they would provide both the information content (this issue) and the visual presentation (#10122) needed for better user understanding of subagent behavior.

[!NOTE] The Plan subagent is particularly affected because it's designed to explore codebases and gather information, often using multiple skills in sequence. Without visibility into skill usage, users are left wondering what the agent is searching for or why it's taking certain actions.