The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill facilitates the installation and execution of third-party code from external sources via npx skills add <package>. It specifically recommends using the -y flag, which bypasses confirmation prompts and allows the agent to execute remote code automatically.
[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands using npx, including npx skills find, npx skills add, and npx skills init. This provides a direct interface for running system commands based on external or user-provided input.
[EXTERNAL_DOWNLOADS]: The skill relies on downloading packages and content from external sources such as GitHub and skills.sh. While some sources like vercel-labs are well-known, the skill allows for the installation of packages from any unverified GitHub repository.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8).
Ingestion points: Untrusted data enters the agent context through search results returned by npx skills find [query], which fetches metadata (names, descriptions) from the external open skills ecosystem.
Boundary markers: The instructions lack explicit boundary markers or instructions to ignore embedded commands within the search results.
Capability inventory: The agent has the capability to write to the filesystem and execute code via the npx skills add command.
Sanitization: There is no mention of sanitizing or validating the output of the search results before presenting them to the user or acting upon them.

find-skills