The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill facilitates the installation of code from remote repositories. Specifically, it provides instructions for using npx skills add <owner/repo@skill> to download and install skill components directly from arbitrary GitHub repositories.
[COMMAND_EXECUTION]: The skill utilizes the npx utility to execute commands that search for and install packages. The inclusion of the -g -y flags in npx skills add <package> -g -y allows for global installation without user intervention, which executes the package's setup scripts on the system.
[EXTERNAL_DOWNLOADS]: The skill's workflow involves fetching multiple file types, including scripts and markdown files, from external domains and saving them to the local filesystem at ~/.claude/skills/.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from external sources which creates a surface for indirect prompt injection attacks.
Ingestion points: Data retrieved from external searches and skill lookups via the prompts.chat service and GitHub repositories.
Boundary markers: The instructions do not define delimiters or warnings to prevent the agent from following instructions embedded within the external skills or prompts.
Capability inventory: The skill has the ability to write files to the system and execute shell commands through the npx package manager.
Sanitization: There are no mentioned procedures for validating or sanitizing the content of the third-party skills before they are installed and utilized.

prompts-chat