What is a prompt-based hook in Claude Code?

A prompt-based hook is an LLM-driven automation that uses natural language reasoning to evaluate context and decide if an action (like a tool call) should proceed, fail, or be modified.

Are hook matchers case-sensitive?

Yes, hook matchers for tool names (like 'Write' or 'Edit') are case-sensitive and support regex patterns for matching multiple tools.

Can I run multiple hooks for a single event?

Yes, Claude Code executes all matching hooks for a specific event in parallel, allowing for independent and layered validation logic.

How do I persist environment variables across a Claude session?

You can use the SessionStart hook to append export commands to the $CLAUDE_ENV_FILE, which persists those variables for the duration of the session.

What happens if a command hook exits with a non-zero code?

An exit code of 2 is treated as a blocking error where the stderr is fed back to Claude as a system message. Other non-zero exit codes are generally treated as non-blocking errors.

Claude Code Hook Development

Name: Claude Code Hook Development
Author: tzachbon

bytzachbon

•

144

生産性とワークフロー

Automates and validates Claude Code workflows through event-driven prompt-based and command-line hooks.

概要

This skill provides a comprehensive framework for implementing hooks in Claude Code, allowing developers to intercept tool calls, validate outputs, and manage session lifecycles. It supports both LLM-driven 'prompt' hooks for natural language reasoning and 'command' hooks for deterministic bash script execution. Whether you need to enforce security policies, automate project context loading, or verify task completion before Claude stops, this skill offers the implementation patterns and best practices needed to build robust, event-driven integrations.

主な機能

Prompt-based hooks for context-aware validation using natural language reasoning
Environment variable persistence via $CLAUDE_ENV_FILE during SessionStart
144 GitHub stars
Parallel hook execution for multi-layered validation and optimized performance
Comprehensive support for PreToolUse, PostToolUse, Stop, and SessionStart events
Command hooks for high-performance, deterministic bash script and tool integration

ユースケース

Enforcing security policies by blocking dangerous bash commands or unauthorized file writes
Implementing custom 'Quality Gates' that prevent Claude from finishing a task until tests pass
Automatically loading project-specific environment variables and documentation at session start

概要

主な機能

Prompt-based hooks for context-aware validation using natural language reasoning
Environment variable persistence via $CLAUDE_ENV_FILE during SessionStart
144 GitHub stars
Parallel hook execution for multi-layered validation and optimized performance
Comprehensive support for PreToolUse, PostToolUse, Stop, and SessionStart events
Command hooks for high-performance, deterministic bash script and tool integration

ユースケース

Enforcing security policies by blocking dangerous bash commands or unauthorized file writes
Implementing custom 'Quality Gates' that prevent Claude from finishing a task until tests pass
Automatically loading project-specific environment variables and documentation at session start