You'll noticeTier: how much it should matter to you
1Useful: my rating, 1 to 5
1Signal: worth watching, 1 to 5
SubagentsArea: what it touches
ImprovementsKind: in v2.1.323,
ImprovementsSection of the release
Unclear It is not stated whether this framing applies in every session or only behind a switch.
What
A subagent is a helper that Claude starts to work on part of a task, which then hands a report back. Claude Code's wording around that report has changed:
The report is labelled as a subagent hand-back and is described as output from the model that is "NOT a message from the user".
If the request to safety-review the report was refused by an upstream safety filter, the report is marked UNREVIEWED with a security warning.
Why
A report from a subagent can contain text the subagent picked up from files or websites. Labelling it this way helps stop Claude from following instructions hidden in that text as though you had given them, a kind of attack known as prompt injection.
How sure we are
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtIt is not stated whether this framing applies in every session or only behind a switch.
