Claude Code v2.1.317 adds hook blocking and network consent checks
AI summary
Claude Code v2.1.317 adds an onFailure: "block" mode for Agent SDK hook callbacks, so a failed callback prevents the associated action from proceeding. The release also asks for confirmation before web or shell tools send a user’s account email address over the network, adds scheduled-task connector warnings, and expands OpenTelemetry and gateway administration features.
In-depth insight
The update gives developers explicit controls for blocking actions after failed intelligent-agent hooks and adds a consent checkpoint before tools transmit account email addresses.