hook-development

Claude Code hooks validate tools and automate workflows by defining event-driven configurations in hooks.json. You can apply prompt-based or command-based hooks across PreToolUse, PostToolUse, Stop, and SessionStart events to enforce policy and coordinate automated handoffs.

Prompt-based hooks in Claude Code use prompt instructions to guide behavior, while command-based hooks execute shell commands for direct processing. This skill provides patterns for both approaches to handle diverse events like loading context at session start or enforcing validation on tool calls.

Claude Code hooks achieve portability by using the ${CLAUDE_PLUGIN_ROOT} variable for paths instead of hardcoded locations. This skill provides configuration patterns and best practices to ensure reliable operation across different environments and setups.

Yes, jq is required as a dependency for this Claude Code hooks skill. It processes JSON data within command-based hooks to parse event payloads, validate tool calls, and coordinate automated handoffs across PreToolUse and PostToolUse events.

You test Claude Code hooks by defining configurations in hooks.json and verifying them with the provided test-hook utility. This skill includes testing guidance and debugging patterns to confirm correct triggering and outcomes across session startup and runtime maintenance.

Claude Code hooks require careful timeout and configuration management to avoid blocking workflows. This skill addresses limitations by providing best practices for secure operation, but complex cross-event automation may need manual coordination and testing to ensure reliability.