Hook Development for Claude Code FAQs

Question 1

Can I persist environment variables across a Claude session?

Accepted Answer

Yes, by using a SessionStart command hook, you can write export statements to the path stored in the $CLAUDE_ENV_FILE environment variable.

Question 2

How can I block dangerous commands using this skill?

Accepted Answer

You can implement a PreToolUse hook with a regex matcher for tools like 'Bash' or 'Write' and use a prompt or script to deny execution if it violates security policies.

Question 3

What is the difference between the plugin hooks.json and user settings.json formats?

Accepted Answer

Plugin-level hooks in hooks/hooks.json require a 'hooks' wrapper object and can include a description, while user settings in .claude/settings.json use a direct top-level event format.

Question 4

What is the difference between prompt-based and command hooks?

Accepted Answer

Prompt-based hooks use natural language reasoning to evaluate context and make flexible decisions, whereas command hooks execute bash scripts for fast, deterministic file system operations and tool integrations.

Question 5

Do hooks run sequentially or in parallel?

Accepted Answer

All matching hooks for a specific event execute in parallel. This design requires hooks to be independent as they do not see each other's outputs during processing.

Hook Development for Claude Code

주요 기능

사용 사례

Hook Development for Claude Code

주요 기능

사용 사례