How do I prevent Claude from running dangerous commands?

You can use a PreToolUse hook with a matcher for 'Bash' to intercept, analyze, and potentially block commands before they are executed.

Can hooks pass information back to Claude?

Yes, hooks can return a systemMessage which is injected into Claude's context, or provide feedback through stdout and specific exit codes.

Where are plugin hooks defined?

Plugin-specific hooks are defined in a hooks/hooks.json file within the plugin directory using a specific wrapper format, while user-level hooks are in .claude/settings.json.

What are the two main types of Claude Code hooks?

Claude Code supports prompt-based hooks for LLM-driven reasoning and command hooks for executing deterministic bash scripts.

What is the difference between Stop and SubagentStop hooks?

The Stop hook triggers when the primary agent considers the task complete, while SubagentStop triggers specifically when a subagent finishes its assigned subtask.

Claude Code Hook Development

Name: Claude Code Hook Development
Author: Claude-Code-Community-Ireland

byClaude-Code-Community-Ireland

•

Developer Tools

Builds event-driven automation and custom workflow logic to validate tool use and enhance Claude Code behavior.

This skill provides a comprehensive framework for implementing lifecycle hooks within Claude Code, allowing developers to intercept and modify tool calls, enforce security policies, and automate repetitive tasks. It covers both high-level prompt-based hooks for complex reasoning and bash-driven command hooks for deterministic checks. By leveraging events like PreToolUse and SessionStart, users can create more robust, context-aware, and secure AI-driven development environments that align with specific project standards.

Key Features

01Deterministic bash command hooks for fast file system operations and tool checks

02Comprehensive support for lifecycle events including PreToolUse, PostToolUse, and Stop

03Implementation of prompt-based hooks for context-aware AI reasoning and validation

04Session management and environment variable persistence via SessionStart and $CLAUDE_ENV_FILE

05Granular tool matching using regex and wildcard patterns for precise event interception

066 GitHub stars

Use Cases

01Automatically loading project-specific context and environment variables at the start of a session

02Enforcing code quality or testing standards before allowing the agent to finalize a task

03Preventing dangerous terminal commands or sensitive file access through PreToolUse validation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add claude-code-community-ireland/claude-code-resources hook-development

For use in Claude.ai and ChatGPT

Download Skill