Where should I store my Hookify rule files?

Rules should be stored in your project's root directory under .claude/hookify.{rule-name}.local.md to ensure they are properly detected and applied to your session.

How do I trigger a rule when a file is edited?

Use the 'file' event type in your rule frontmatter. You can then use conditions to match specific file paths or the content being added (new_text) to trigger warnings for things like debug logs or security risks.

Can I prevent Claude from running specific bash commands?

Yes, by creating a rule with the 'bash' event and a regex pattern for the command, you can set the action to 'block' to prevent Claude from executing dangerous or unauthorized commands.

What are Hookify rules in Claude Code?

Hookify rules are markdown files with YAML frontmatter stored in the .claude directory that define patterns Claude should watch for and specific actions (warn or block) to take when those patterns are matched.

Hookify Rule Writer

Name: Hookify Rule Writer
Author: brixtonpham

bybrixtonpham

0•

セキュリティとテスト

Configures custom validation rules and automated guardrails to govern Claude Code's behavior and command execution.

The Hookify Rule Writer skill provides a standardized framework for creating project-specific guardrails within Claude Code. By defining markdown-based rules with YAML frontmatter, developers can monitor bash commands, file edits, and session completions to enforce security policies, coding standards, and workflow checklists. This skill allows users to programmatically prevent dangerous operations like accidental deletions, flag debug code before it reaches production, and ensure that mandatory steps—such as running tests—are completed before an agent finishes its task.

主な機能

010 GitHub stars

02Support for dynamic rule loading through .local.md file organization

03Define multi-condition rules using fields like file_path, new_text, and command

04Create regex-based triggers for bash commands and file system operations

05Implement session-end checklists triggered by the 'stop' event

06Configure 'warn' or 'block' actions to enforce safety and compliance

ユースケース

01Preventing the execution of dangerous shell commands like 'rm -rf' or 'chmod 777'

02Enforcing coding standards by warning against console logs, eval(), or hardcoded API keys

03Creating mandatory pre-flight checklists that trigger before Claude exits a session

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add brixtonpham/dot-claude writing-rules

For use in Claude.ai and ChatGPT

Download Skill