[FEATURE] VSCode Extension: Missing @agent syntax support and unclear agent call visualization

Preflight Checklist

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

Problem Statement

What problem are you trying to solve?

I cannot easily discover and invoke agents in the VSCode extension. The experience differs significantly from the CLI:

• No autocomplete suggestions when typing @agent-name - I don't know what agents are available
• No clear visual indicator when an agent call completes - I can't confirm if an agent was actually invoked
• No visibility into agent details (tool calls, tokens, duration)

Why do I need this feature?

Agents are a core capability I rely on for specialized tasks (code review, planning, debugging, security analysis). I work in VSCode daily and frequently switch between different agents based on the task at hand.

The missing autocomplete makes agent discovery difficult:

• I have to remember exact agent names or check my ~/.claude/agents/ directory manually
• I cannot quickly browse available agents while composing a prompt
• New users won't discover agents at all without reading documentation

The unclear visualization creates uncertainty:

• Results appear as plain text with no visual distinction from normal responses
• I cannot see tool usage, token costs, or agent reasoning
• I'm left uncertain: did the agent actually run?

Help us understand your workflow

My typical workflow in CLI:

1. I'm working on Python code and want specialized review
2. I type @pyth in the prompt
3. CLI immediately shows autocomplete suggestions:

@python-reviewer (Python code review specialist)
@python-patterns (Python design patterns guide)
...

1. I select python-reviewer from the list
2. After execution, I see clear output:

python-reviewer(Python code review)
⎿ Done (5 tool uses · 35k tokens · 42s)
(ctrl+o to expand

1. I can see exactly what the agent did, expand for details, and continue

My workflow in VSCode extension:

1. I'm working on Python code in VSCode
2. I type @pyth expecting agent suggestions
3. Nothing appears - no autocomplete, no agent list
4. I have to remember the exact name and type the full request manually
5. If I don't know the agent name, I can't use it
6. After the response, there's no visual indicator an agent was used
7. I cannot see tool usage, tokens, or expand for details

This makes agent-based workflows impractical in VSCode, forcing me to switch to CLI when I need specialized agents.

Proposed Solution

Proposed Solution

Make the VSCode extension agent experience consistent with CLI:

1. Agent autocomplete suggestions:

• When typing @, show a dropdown list of available agents
• Filter the list as user continues typing (e.g., @pyth shows python-related agents)
• Each suggestion displays agent name and brief description
• User can select from the list or continue typing

2. Clear agent call visualization:

• Display agent name and description as a header (e.g., python-reviewer (Python code review))
• Show completion summary: tool uses, tokens, duration
• Provide expandable output section (similar to CLI's ctrl+o to expand)
• Visually distinguish agent responses from normal assistant responses

Ideal user experience in VSCode:

1. Type @pyth in prompt
2. Dropdown appears with filtered agent suggestions
3. Select python-reviewer from list
4. Agent executes and displays:

Alternative Solutions

Alternative Solutions

Current workaround:

• Manually type full agent request: "use python-reviewer to review this code"
• Check ~/.claude/agents/ directory to see available agents
• Switch to CLI when I need agent-based workflows

Limitations of workaround:

• Requires memorizing agent names
• No visibility into agent execution
• Forces context switching between VSCode and CLI
• Inefficient for frequent agent usage

Alternative considered:

• Could display available agents in a sidebar panel
• However, inline autocomplete is more natural and matches CLI behavior
• Sidebar approach would still require extra navigation steps

Priority

High - Significant impact on productivity

Feature Category

Interactive mode (TUI)

Use Case Example

My typical workflow in CLI:

1. I'm working on Python code and want specialized review
2. I type @pyth in the prompt
3. CLI immediately shows autocomplete suggestions:

@python-reviewer (Python code review specialist)
@python-patterns (Python design patterns guide)
...

1. I select python-reviewer from the list
2. After execution, I see clear output:

python-reviewer(Python code review)
⎿ Done (5 tool uses · 35k tokens · 42s)
(ctrl+o to expand

1. I can see exactly what the agent did, expand for details, and continue

My workflow in VSCode extension:

1. I'm working on Python code in VSCode
2. I type @pyth expecting agent suggestions
3. Nothing appears - no autocomplete, no agent list
4. I have to remember the exact name and type the full request manually
5. If I don't know the agent name, I can't use it
6. After the response, there's no visual indicator an agent was used
7. I cannot see tool usage, tokens, or expand for details

Additional Context

_No response_