The Agent Skills Directory

[NO_CODE]: The skill package contains no executable scripts, configuration files (like package.json or requirements.txt), or binary assets. It consists entirely of documentation and metadata in the SKILL.md file.
[EXTERNAL_DOWNLOADS]: The skill description specifies its primary purpose is to facilitate downloading agent skills from "arbitrary GitHub repos." This identifies an external download capability inherent to the tool the skill wraps.
[REMOTE_CODE_EXECUTION]: The intended functionality involves installing and running third-party code from the internet. While the skill itself provides no code to perform these actions, the capability it describes represents a potential vector for remote code execution if used with untrusted sources.
[PROMPT_INJECTION]: The skill defines a surface for the ingestion of external instructions by facilitating the installation of third-party skills.
Ingestion points: GitHub repositories specified by the user or agent at runtime.
Boundary markers: None are defined in the skill's metadata to differentiate between trusted and untrusted inputs.
Capability inventory: The skill facilitates the installation of additional functional capabilities (skills) into the agent's environment.
Sanitization: No sanitization or validation logic is present in this documentation-only skill package.

add-skill