Automates and controls Claude Code behavior using event-driven shell commands and LLM-evaluated prompt hooks.
This skill provides a comprehensive framework for implementing programmatic control over Claude Code's lifecycle through event-driven automation. It enables developers to configure hooks that fire during specific events—such as tool usage, session starts, or user submissions—allowing for automated code formatting, security validation, and context injection. By mastering the hierarchy of matchers and hook types, users can create sophisticated workflows that block destructive actions, log CLI interactions, and integrate external tools directly into their AI development environment without modifying core code.
Key Features
01Dual-mode execution supporting both shell commands and LLM-evaluated prompts
02Blocking capabilities for command validation and safety enforcement
03Advanced tool filtering using regex matchers and MCP server patterns
04Context injection and auto-formatting for streamlined development workflows
05Event-driven automation for tool usage, session lifecycle, and user interactions
060 GitHub stars
Use Cases
01Automating post-edit workflows such as running linters or code formatters after file changes
02Injecting custom project metadata or sprint-specific context into every new Claude session
03Enforcing safety protocols by blocking destructive commands like force pushes or recursive deletes
