create-hooks

Claude Code hooks automate workflows and validate commands by triggering custom scripts or LLM prompts during tool usage, session events, or user interactions. You can configure regex matchers to target specific tools and implement safety checks to prevent harmful behavior.

Command hooks execute simple shell commands for straightforward automation, while prompt hooks leverage LLM prompts to handle complex logic and decision-making. Both hook types apply regex pattern matching to match specific tools or tool categories before triggering.

Yes, jq is required as a dependency to configure event-driven hooks in Claude Code. You also need knowledge of shell commands and LLM prompts to implement the hook logic effectively for automated workflows and command validation.

To set up a Bash command logging hook, use the create-hooks guidance to configure a hook for tool usage events. Apply a regex matcher to target the Bash tool category, ensuring the hook triggers and logs every shell command executed during your Claude Code session.

Yes, you can implement security checks and validation with Claude Code hooks to prevent malicious or harmful behavior. By configuring hooks with regex patterns and prompt-based logic, you can intercept and validate tool usage before execution.

Use regex matchers when you need to trigger hooks selectively for specific tools or tool categories. This pattern matching capability allows you to apply automation and validation rules only to targeted events, preventing unnecessary execution on unrelated session interactions.