What
A subagent is a helper Claude starts to work on part of a task. A PermissionRequest hook is a script you can set up to run when a tool call needs a permission decision. When a permission answer is passed on from a subagent and that answer is a denial, Claude Code now skips these hooks. Before, they always ran.
Two smaller changes come with it:
- Claude Code records whether the prompt held back, offered a broad, or offered an exact "always allow" option.
- If a hook fails after the session has been stopped, the error is passed on rather than ignored.
Why
If you rely on PermissionRequest hooks, they will not fire for denials handed up from a subagent.
Names in the bundlePermissionRequest
The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.
Added since
A small documentation edit on Run Claude Code programmatically touched a line naming PermissionRequest after this was published.
Calls to tools you didn't list run if they need no approval. In a [Manual mode](/docs/en/permission-modes#available-modes) `claude -p` run, a call that needs approval is denied unless something approves it, such as a [`PermissionRequest` h…headless see the edit
Confirmed since
Anthropic's documentation has since written up PermissionRequest, on Error reference.
| `Managed settings contain invalid entries, including PreToolUse/PermissionRequest hooks that are NOT loaded; fix or remove each entry:` | [Configuration warnings](#managed-settings-hooks-not-loaded) |errors see the edit
Two sources agreeTwo things we can check say the same as this entry.
Anthropic's documentation agrees
Anthropic's documentation has since written up PermissionRequest, on Error reference.
Anthropic's release notes agree
Changed auto mode in Cowork cloud sessions: a PermissionRequest hook can no longer approve shell commands or subagents; those prompts wait…