What are the two types of hooks supported?

Claude Code supports prompt-based hooks, which use LLM reasoning for complex validation, and command hooks, which use bash scripts for fast, deterministic checks.

Can I run hooks for specific tools only?

Yes, you can use matchers—including exact names, multiple tool names, or regex patterns—to trigger hooks only for specific operations like Write or Edit.

How do hooks handle errors?

Hooks use specific exit codes: 0 for success, 1 for non-blocking errors, and 2 for blocking errors that halt the entire operation for safety.

Can hooks be used to modify tool inputs?

Yes, PreToolUse hooks can be used for validation or modification, allowing you to intercept and adjust data before a tool is executed.

Are environment variables available inside hooks?

Yes, hooks have access to variables like $CLAUDE_PROJECT_DIR and $CLAUDE_PLUGIN_ROOT, ensuring portability and context awareness.

Claude Code Hook Management

Name: Claude Code Hook Management
Author: Git-Fg

byGit-Fg

•

개발자 도구

Facilitates the creation, configuration, and implementation of event-driven hooks to customize Claude Code's tool execution and session lifecycle.

소개

This skill provides a comprehensive framework for managing Claude Code hooks, enabling developers to inject custom logic before, during, or after tool usage. It covers both prompt-based LLM validation and command-driven bash scripts, allowing for complex safety checks, automated logging, and task verification. By standardizing the implementation of events like PreToolUse and Stop, it ensures that Claude Code operations remain secure, performant, and aligned with project-specific constraints, providing granular control over how the AI interacts with the local environment.

주요 기능

Support for both LLM-driven prompt hooks and deterministic bash command hooks
Advanced matching patterns using regex and wildcards for granular tool control
Built-in security best practices for validating file operations and external inputs
Comprehensive event lifecycle management including PreToolUse, PostToolUse, and Session events
1 GitHub stars
Standardized JSON-based input/output protocols for seamless integration

사용 사례

Verifying task completion status before the agent terminates a session
Automating code validation or linting immediately after Claude edits a file
Implementing safety checks to prevent unauthorized file deletions or sensitive data access

소개

주요 기능

Support for both LLM-driven prompt hooks and deterministic bash command hooks
Advanced matching patterns using regex and wildcards for granular tool control
Built-in security best practices for validating file operations and external inputs
Comprehensive event lifecycle management including PreToolUse, PostToolUse, and Session events
1 GitHub stars
Standardized JSON-based input/output protocols for seamless integration

사용 사례

Verifying task completion status before the agent terminates a session
Automating code validation or linting immediately after Claude edits a file
Implementing safety checks to prevent unauthorized file deletions or sensitive data access