The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill facilitates the discovery and installation of third-party code from GitHub using the npx skills add command. Since these 'skills' are modular packages that extend agent capabilities with tools and scripts, this represents a remote code execution vector.\n- [EXTERNAL_DOWNLOADS]: The skill is designed to fetch content from GitHub repositories and the skills.sh domain. While it mentions well-known organizations like Vercel, the mechanism supports installation from any arbitrary GitHub account.\n- [COMMAND_EXECUTION]: The instructions direct the agent to execute shell commands using npx, specifically npx skills find for discovery and npx skills add for installation.\n- [PRIVILEGE_ESCALATION]: The skill explicitly instructs the agent to use the -y flag when installing packages (e.g., npx skills add <package> -g -y). This flag suppresses interactive confirmation prompts, allowing the agent to install external software automatically without a manual 'yes/no' approval step from the user.\n- [INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by processing and displaying untrusted data from an external ecosystem.\n
Ingestion points: Data enters the agent's context through the output of the npx skills find command (SKILL.md).\n
Boundary markers: No delimiters or warnings are used to isolate search results from the agent's instructions.\n
Capability inventory: The agent has the capability to execute shell commands and install software based on the ingested data.\n
Sanitization: There is no evidence of validating or sanitizing the search results before they are presented or acted upon.

find-skills