[DOCS] Subagent docs don't explain message routing when viewing a running subagent

Documentation Type

Missing documentation (feature not documented)

Documentation Location

https://code.claude.com/docs/en/sub-agents

Section/Topic

/agents Running tab, subagent pane behavior, and how typed messages are attributed and recorded when interacting with a running subagent

Current Documentation

The docs currently say:

The /agents command opens a tabbed interface for managing subagents. The Running tab shows live subagents and lets you open or stop them.

And the desktop docs say:

The tasks pane shows the background work running inside the current session: subagents, background shell commands, and workflows. Click any entry to see its output in the subagent pane or stop it.

No documentation currently explains what happens if you type a message while a running subagent instance is open, whether that message is sent to the subagent or the parent conversation, or which transcript should contain that message.

What's Wrong or Missing?

Changelog v2.1.113 includes this fix:

Fixed messages typed while viewing a running subagent being hidden from its transcript and misattributed to the parent AI

That changelog entry exposes a user-facing behavior that the docs do not describe: when a user is viewing a running subagent instance and types a message, that message belongs to the subagent conversation and should appear in the subagent's transcript rather than being attributed to the parent AI.

Current docs explain how to open a running subagent and view its output, but they do not explain:

• whether typing in that view sends the message to the running subagent
• how that differs from typing into the main conversation to steer the parent session
• where the resulting message is recorded in transcript history

Without that explanation, users cannot reliably understand which agent received their instruction or where to look for the resulting conversation history.

Suggested Improvement

Add a short subsection to the subagents docs, with a matching note in the desktop tasks/subagent pane docs, that explains the interaction model for running subagents.

Suggested content:

• Opening a running subagent instance lets you interact with that specific subagent while it is running.
• Messages typed in that view are appended to the subagent conversation, not attributed to the parent AI.
• Those messages should appear in the subagent's transcript/history.
• If useful, link this behavior to the existing section about subagent transcripts and persistence so users know where that history lives.

This would make the running-subagent UI and transcript model explicit instead of leaving users to infer it from behavior.

Impact

Medium - Makes feature difficult to understand

Additional Context

Affected Pages:

| Page | Context |
|------|---------|
| https://code.claude.com/docs/en/sub-agents | /agents Running tab and subagent transcript behavior |
| https://code.claude.com/docs/en/desktop | Tasks pane and subagent pane behavior when viewing running background work |

Total scope: 2 pages affected

Source: Changelog v2.1.113

Exact changelog entry: Fixed messages typed while viewing a running subagent being hidden from its transcript and misattributed to the parent AI