The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill facilitates the download and installation of modular packages from remote GitHub repositories using the 'npx skills add' command, allowing arbitrary third-party code to be executed in the agent's environment.
[COMMAND_EXECUTION]: The instructions explicitly direct the agent to use the '-y' flag when installing skills, which bypasses safety confirmation prompts and allows the agent to install and execute new code without user oversight.
[EXTERNAL_DOWNLOADS]: The skill retrieves capability configurations and code from 'skills.sh' and various GitHub repositories. While it references well-known organizations like Vercel, it supports installing from any arbitrary repository identifier based on user or automated queries.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. It ingests data from an external ecosystem (the 'npx skills find' search results) and possesses high-risk capabilities like code installation ('npx skills add'). There are no boundary markers or sanitization processes described to prevent malicious skill descriptions from influencing the agent's behavior or tricking it into installing harmful software.

find-skills