The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill metadata and instructions direct the agent to install the clawhub package from the global npm registry.
[EXTERNAL_DOWNLOADS]: The skill communicates with the remote registry at https://clawhub.com to fetch, search, and update skill content.
[COMMAND_EXECUTION]: The skill utilizes the clawhub CLI to execute lifecycle commands such as install, update, publish, and login on the host system.
[PROMPT_INJECTION]: The skill facilitates the installation of third-party agent skills from the ClawHub registry, which introduces an attack surface for indirect prompt injection.
Ingestion points: Untrusted data enters the context through clawhub search results and clawhub install packages.
Boundary markers: Absent. There are no instructions to the agent to treat registry content as untrusted or to use delimiters.
Capability inventory: The skill provides capabilities to write new files to the skills/ directory and execute them via the agent platform.
Sanitization: The instructions do not specify any validation, integrity checking, or sanitization of the content retrieved from the registry.

clawhub