[BUG] VSCode extension: long slash-command/skill arguments are not collapsed with SHOW MORE/LESS toggle

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

In the Claude Code VS Code extension, long regular user messages are automatically collapsed with a SHOW MORE / SHOW LESS toggle — this works correctly.

However, when a slash command (skill invocation) is sent with a long argument, e.g.:

/skill-creator
<long multi-line content, dozens/hundreds of lines>
...

the message is not collapsed at all — the SHOW MORE / SHOW LESS toggle never appears, and the entire argument is rendered expanded in the conversation. This pushes Claude's subsequent output far below the fold and makes it practically impossible to read the assistant's response without scrolling through the whole argument block every time.

So the collapse UI appears to be applied only to plain user messages, and not to user messages whose first line is a slash command with a multi-line trailing argument.

Steps to Reproduce

1. Open the Claude Code VS Code extension.
2. In the prompt input, type a slash command that invokes a skill / custom command, followed by a long multi-line argument on subsequent lines (use Shift+Enter for new lines). Roughly 40+ lines is enough to reliably trigger the issue. Example:

   /skill-creator
   <paste ~50 lines of spec text here>

1. Submit the message.
2. Observe the rendered user message in the conversation view.

Expected Behavior

The long slash-command message should be collapsed with a SHOW MORE / SHOW LESS toggle, the same way a plain long user message is collapsed. The assistant's response should remain visible immediately below without requiring the user to scroll past the full argument.

Actual Behavior

• The slash-command user message is rendered fully expanded.
• No SHOW MORE / SHOW LESS toggle is shown.
• Claude's response is pushed off-screen and is hard to reach.

Environment

• Platform: VS Code extension (Claude Code)
• OS: <fill in, e.g. macOS 14.x / Windows 11 / Ubuntu 22.04>
• VS Code version: <Help → About>
• Claude Code extension version: <fill in>
• Claude Code CLI version (if relevant): <fill in>

Additional Context

• Plain long user messages (without a slash command) are collapsed correctly with the toggle, so the collapse feature itself works — the bug appears to be specific to messages that start with a slash command / skill invocation.
• This is distinct from #39809 (which reports that the SHOW MORE toggle is unresponsive once collapsed). In this bug, the collapse UI itself is never applied.
• Screenshots / short screen recording showing both cases side by side would help — please attach if possible.

What Should Happen?

The long slash-command message should be collapsed with a SHOW MORE / SHOW LESS toggle, the same way a plain long user message is collapsed. The assistant's response should remain visible immediately below without requiring the user to scroll past the full argument.

Error Messages/Logs

Steps to Reproduce

Steps to Reproduce

Baseline (works correctly):

1. Open the Claude Code VS Code extension and start a new chat.
2. In the prompt input, paste ~50 lines of plain text (any content, no leading slash command).
3. Submit.
4. ✅ The message is rendered collapsed with a SHOW MORE toggle; clicking it expands the message, and SHOW LESS collapses it again.

Bug case (collapse UI is missing):

1. In the same (or a fresh) chat, type / to open the slash-command picker and select any skill or custom slash command that accepts arguments — e.g. /skill-creator, or a user command defined at .claude/commands/<name>.md that uses $ARGUMENTS.
2. Press Shift+Enter to start a new line, then paste the same ~50 lines of text used in the baseline as the command's argument. The message should look roughly like:

/skill-creator
<line 1>
<line 2>
...
<line 50>

1. Submit.
2. ❌ The user message is rendered fully expanded with no SHOW MORE / SHOW LESS toggle. The entire argument fills the viewport and Claude's response is pushed far below, requiring manual scrolling past the whole argument every time the message is revisited.

Frequency: 100% reproducible on my setup. The only variable that flips the behavior between Baseline and Bug case is the presence of a leading slash command — same content length, same chat, same session.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.118

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

iTerm2

Additional Information

_No response_