The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the 'npx skills' CLI tool. Specifically, it suggests using the '-y' flag during installation ('npx skills add owner/repo@skill -g -y'), which bypasses confirmation prompts and allows for the automatic installation of external code. It also includes the '-g' flag for global installation.
[EXTERNAL_DOWNLOADS]: The skill is designed to download and install packages from the 'open agent skills ecosystem' via GitHub and 'skills.sh'. While the documentation highlights trusted organizations, the mechanism supports installing arbitrary packages from the public ecosystem.
[REMOTE_CODE_EXECUTION]: By design, the skill facilitates the execution of remote code by installing packages using 'npx', which downloads and executes the package logic.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes data from external search results ('npx skills find [query]') without explicit sanitization or boundary markers. This could allow a malicious package description to influence the agent's behavior. The skill mitigates this risk by providing a verification checklist for the agent (checking install counts, stars, and source reputation) before recommending or installing a skill.

find-skills