The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions to run pnpx skills add <owner/repo@skill> -y, which downloads and installs executable code from external repositories. The use of the -y flag suggests the installation occurs without a manual confirmation step, enabling the execution of unverified remote code in the agent's environment.\n- [EXTERNAL_DOWNLOADS]: The skill interfaces with a third-party registry at https://skills.sh/ to fetch and install external content. This registry is not identified as a trusted organization or well-known service, presenting a supply chain risk.\n- [COMMAND_EXECUTION]: The skill executes shell commands through the pnpx utility to manage packages and interact with the external 'skills' ecosystem.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting and displaying data from an unverified public registry.\n
Ingestion points: Data returned from the pnpx skills find command (SKILL.md).\n
Boundary markers: No delimiters or safety warnings are used when presenting search results to the agent.\n
Capability inventory: The skill can install software and execute shell commands (SKILL.md).\n
Sanitization: There is no evidence of sanitization for the metadata (names and descriptions) retrieved from the external registry.

find-skills