CLI/TUI subagent wait rows show UUIDs instead of nicknames

What version of Codex CLI is running?

codex-cli 0.136.0

What subscription do you have?

Pro20x

Which model were you using?

gpt-5.5 xhigh

What platform is your computer?

24.6.0 arm64 arm

What terminal emulator and version are you using (if applicable)?

iTerm2 3.5.14

Codex doctor report

What issue are you seeing?

The CLI/TUI sometimes renders subagent UUIDs instead of the user-facing nickname.

Example observed output:

``text
Waiting for 2 agents
└ 019e8d20-9db3-7252-a566-464bf6722f03
019e8d20-9e0f-7593-a55a-9e3bb58c2a8b
``
This is inconsistent because spawn/finished/closed rows can show nicknames, while the waiting row falls back to raw IDs. It makes multi-agent output hard to read.

What steps can reproduce the bug?

1. Start Codex CLI with multi-agent support enabled.
2. Ask it to spawn two subagents with user-facing names/roles, for example one explorer and one reviewer.
3. Ask the main agent to wait for both subagents.
4. Observe the Waiting for 2 agents row in the TUI.

In the affected CLI path, the waiting row may render raw subagent UUIDs instead of the nicknames, even though nickname metadata is already available elsewhere in
the session.

What is the expected behavior?

The CLI/TUI should consistently show the user-facing subagent nickname and role when that metadata is available.

For example:

``text
Waiting for 2 agents
└ Rawls [explorer]
Erdos
``

Additional information

I prepared a minimal display-only fix here:

https://github.com/openai/codex/compare/main...47seek:codex:fix-cli-subagent-nicknames

The fix preserves existing nickname/role metadata through app-server collab item state and uses it during TUI rendering.

It does not change:

• subagent lifecycle
• slot/limit behavior
• routing
• spawn/wait/close execution behavior

Validation:

• cargo test -p codex-app-server-protocol passed
• targeted TUI wait-begin rendering test passed
• local CLI test confirmed Waiting for 2 agents shows names instead of UUIDs