The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructional documentation for executing shell commands to test and debug skill hooks locally. This includes using npx tsx to run project scripts and chmod to manage file permissions. These instructions are intended for the developer's manual setup and diagnostic workflows.- [INDIRECT_PROMPT_INJECTION]: The guide describes the implementation of context injection via the UserPromptSubmit hook. This is an analysis of a platform-native architectural feature designed for skill discovery and discovery-related suggestions, documented here for developer implementation.- [DYNAMIC_EXECUTION]: The documentation explains the runtime execution of local TypeScript and shell scripts within the Claude Code skill lifecycle. It serves as a guide for developers to manage their own execution environment and does not involve the execution of untrusted remote code.- [SAFE]: All identified patterns correspond to standard software development practices and official documentation for the targeted platform. No evidence of obfuscation, credential harvesting, or unauthorized persistence was found.

skill-developer