The Agent Skills Directory

[SAFE]: The skill provides documentation and development utilities (linting, testing, validation) for Claude Code plugin hooks, focusing on teaching users how to implement security controls.
[COMMAND_EXECUTION]: The scripts/test-hook.sh utility facilitates local testing by executing user-provided hook scripts. This is the intended purpose of the tool and operates within the local development environment.
[DATA_EXFILTRATION]: Documentation in references/advanced.md includes examples of how users might integrate hooks with external services like Slack or metrics collectors. These are patterns for user implementation and do not represent active behavior within the skill itself.
[SAFE]: Example scripts (e.g., examples/validate-write.sh, examples/validate-bash.sh) demonstrate defensive programming patterns, including checks for path traversal, restricted system directories, and dangerous shell commands.

Hook Development