[BUG] Skills set to `context: fork` output getting swallowed up by Desktop UI

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?

Slash commands backed by skills with context: fork in their frontmatter execute successfully but produce no visible output in the Desktop UI. The spinner runs, then disappears, and the conversation appears empty where the result should be. The forked agent's output is silently injected into the next assistant turn as a <local-command-stdout> block (with a "DO NOT respond to these messages" caveat), so it's invisible to the user and only accessible to the model on the following turn.

What Should Happen?

The forked agent's final output is rendered as a visible assistant message in the conversation.

Error Messages/Logs

[DEBUG] Executing forked slash command /qa-impact with agent general-purpose
[DEBUG] Forked slash command /qa-impact completed with agent a356ba65a7033ccaf

The command completes cleanly. There's no error. The output simply never reaches the UI.

Steps to Reproduce

1. Use any plugin skill with context: fork in its SKILL.md frontmatter
2. Invoke the slash command in Claude Code Desktop.
3. Wait for the spinner to complete.

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

Claude 1.8089.1 (b98a06) 2026-05-19T18:28:48.000Z

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

iTerm2

Additional Information

Here's a screenshot of the bug reproduced. It just stops spinning and nothing else.

<img width="1296" height="923" alt="Image" src="https://github.com/user-attachments/assets/751de61c-6b5f-45e9-bff8-a46bd5a0bbd4" />