Claude Code Hook Design Architect FAQs

Question 1

What does exit code 2 signify in Claude Code hooks?

Accepted Answer

Exit code 2 acts as a blocking error that stops execution and feeds the stderr content back to Claude or the user to handle a critical failure or security violation.

Question 2

What is the primary difference between a hook and a prompt?

Accepted Answer

Hooks are deterministic, meaning they execute every single time a specific event occurs, whereas prompts are probabilistic and rely on Claude's judgment to follow instructions.

Question 3

What are matchers used for in hook configuration?

Accepted Answer

Matchers use exact strings or regex patterns to define which specific tools (like 'Write' or 'Edit') should trigger PreToolUse and PostToolUse hooks.

Question 4

When should I use a Python script instead of Bash for a hook?

Accepted Answer

Use Python when you need complex validation, JSON parsing from stdin, or advanced logic requiring external libraries via UV for dependency management.

Question 5

How do I communicate with both the user and Claude in a single hook?

Accepted Answer

By returning a JSON object containing 'systemMessage' for user-facing terminal output and 'additionalContext' within 'hookSpecificOutput' for injecting context into Claude's memory.

Claude Code Hook Design Architect

关于

主要功能

使用场景

Claude Code Hook Design Architect

关于

主要功能

使用场景