hook-development

Claude Code hooks are governed by defining structured JSON schemas for events like PreToolUse and Stop, then validating them with linter scripts to ensure safe execution. This Skill provides templates, validation scripts, and debugging guidance for reliable plugin development.

Supported Claude Code hook events include PreToolUse, PostToolUse, Stop, and SessionStart. The Skill provides ready-to-use templates and best practices for implementing both prompt-based and command hooks across these full lifecycle events.

You validate a hooks.json configuration using built-in shell scripts like validate-hook-schema.sh and hook-linter.sh. These tools check JSON schemas, verify portable path references, and ensure secure hook execution before deployment.

Yes, jq is required as a dependency for Claude Code hook development. It is used within the provided validation, testing, and linting scripts to parse and validate JSON structures during hook execution and testing workflows.

Yes, you can use prompt-based hooks as an alternative to command hooks for Claude Code workflow automation. The Skill offers ready-to-use templates and patterns for both approaches, including specific workflows for SessionStart and Stop events.

You test Claude Code hooks using the included test-hook.sh script alongside sample test inputs. This allows you to validate hook behavior, debug execution issues, and verify safe, auditable hook performance before real-world deployment.