The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the npx skills CLI tool to perform searches and installations of modular agent packages. It includes instructions for the agent to execute shell commands such as npx skills find and npx skills add with the -g (global) and -y (yes) flags.
[EXTERNAL_DOWNLOADS]: The skill is explicitly designed to download and install third-party code from the skills.sh ecosystem and GitHub repositories, including references to trusted sources like Vercel Labs.
[PROMPT_INJECTION]: As an entry point for external code, the skill introduces a surface for indirect prompt injection. Malicious instructions could be embedded within the description or logic of third-party skills discovered and installed through this process.
Ingestion points: User-provided search queries used as arguments for the find command and the names of packages to be installed.
Boundary markers: The instructions do not specify sanitization or delimiters for the user query when interpolated into the shell command.
Capability inventory: The skill allows for shell command execution and installation of arbitrary packages via npx.
Sanitization: No explicit sanitization or validation of the search query or package source is mentioned in the provided instructions.

find-skills