Unclear What the extra processing step does to hook text is not stated.
What
Hooks (your own scripts that Claude Code runs at set points) can hand text back to the model through systemMessage and hookSpecificOutput.additionalContext. That text now goes through an extra processing step before it is added to the conversation. Before, it was added exactly as the hook returned it.
Why
The way hook text appears to the model may be different, so check any hook that depends on its exact wording.
Names in the bundlesystemMessagehookSpecificOutput.additionalContext
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.
Confirmed since
Anthropic's documentation has since written up systemMessage, on Hooks reference.
Claude Code discards `reason`, `systemMessage`, and `continue` from an Elicitation hook's JSON output, whichever `action` you return.hooks see the edit
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubt
What the extra processing step does to hook text is not stated.
Anthropic's documentation agrees
Anthropic's documentation has since written up systemMessage, on Hooks reference.