The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides utility scripts such as hook-linter.sh and test-hook.sh that facilitate the validation and execution of local shell scripts. This is part of the intended developer workflow for testing hook logic. \n- [INDIRECT_PROMPT_INJECTION]: The documentation describes how hooks process external data like user prompts and tool inputs. It includes specific security patterns to mitigate injection risks, such as path traversal checks and destructive operation blocking. \n- [DATA_EXFILTRATION]: Reference material illustrates integration patterns for sending data to external services (e.g., Slack webhooks or remote databases) for logging and notifications. These are presented as user-configured extensions. \n- [DYNAMIC_EXECUTION]: The test-hook.sh script enables the testing of local hook scripts, and the "prompt-based hooks" API allows for dynamic model-based evaluation of tool calls.

hook-development