The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill facilitates the installation of external code packages using npx skills add. This mechanism allows the agent to fetch and integrate modular extensions from remote sources, such as GitHub repositories, into its execution environment.- [COMMAND_EXECUTION]: The instructions include shell commands like npx skills add -g -y. The inclusion of the -y flag is intended to bypass confirmation prompts, which facilitates autonomous installation but reduces the level of user oversight during the package addition process.- [EXTERNAL_DOWNLOADS]: The skill references and downloads content from external sources, including the skills.sh domain and various GitHub repositories. It specifically notes reputable sources like the official GitHub repository for Vercel Labs.- [INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface by ingesting search results from the skills.sh registry via the npx skills find command. These ingestion points in the SKILL.md file present a risk as the skill lacks explicit boundary markers or instructions to ignore potential commands embedded in package descriptions. The skill's capability inventory includes shell execution and global package installation, which could be exploited if malicious metadata is ingested. No sanitization or integrity checking of the external search results is specified before they are presented to the user for potential installation.

find-skills