claude-hooks_skill

Overview

This skill configures and manages Claude Code hooks to validate tools, log activity, send notifications, and run custom automations. It provides patterns and templates to add PreToolUse, PostToolUse, Notification, and Stop hooks quickly and safely. Use it to enforce policies, capture audit trails, and automate cleanup or alerts.

How this skill works

The skill generates and updates hook definitions stored in the CLAUDE configuration file and supplies ready-to-use templates for common events. It wires PreToolUse hooks for input validation and blocking, PostToolUse hooks for logging and cleanup, Notification hooks for alerts, and Stop hooks for final reporting. It also includes matchers and examples to scope hooks to specific tools or contexts.

When to use it

• Validate tool inputs or block unsafe executions
• Capture audit logs and metadata after tool runs
• Send alerts or notifications on specific events
• Automate cleanup, retries, or follow-up tasks
• Enforce security controls or policy checks

Best practices

• Keep hook execution under 1 second to avoid slowing workflows
• Scope hooks with specific matchers to limit blast radius
• Handle errors gracefully and fail-safe (don’t block critical flows)
• Test hooks independently before deploying to production
• Log structured, minimal output to simplify auditing

Example use cases

• PreToolUse validation that blocks sensitive data exfiltration attempts
• PostToolUse logger that records tool outputs and runtime metadata for audits
• Notification hook that sends Slack or email alerts when a failure pattern appears
• Stop hook that compiles a final run report and posts to a dashboard
• Automation hook that retries flaky tools or triggers downstream jobs