The Agent Skills Directory

[SAFE]: The skill provides a collection of templates and examples for Claude Code hooks, which are event-driven automations for the Claude Code environment.
[COMMAND_EXECUTION]: The skill contains scripts designed to execute development tools such as linters (eslint, ruff), formatters (prettier, black), and testers (pytest, jest). These are standard development automations and include explicit logic to prevent dangerous command execution through security gates and validation patterns.
[DATA_EXFILTRATION]: A notification forwarder script is provided that can send data to Slack or Discord webhooks. This functionality is clearly documented, relies on user-provided environment variables for configuration, and is intended for legitimate workflow notifications. No hardcoded credentials or unauthorized data transmission logic was found.
[SECURITY_PRACTICES]: The skill includes extensive documentation on security best practices for hooks, including path sanitization, secret scanning (detecting AWS/GitHub keys), and preventing command injection through proper variable quoting. The scripts provided as examples (e.g., file-protector.py, path-sanitizer.py) are defensive tools designed to enhance the security of the agent's environment.

hooks-builder