The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute vaxis skills get core in the shell to retrieve its operating instructions. This pattern shifts the skill's logic from a static file to the dynamic output of a local command.\n- [PROMPT_INJECTION]: The skill implements an indirect instruction injection surface by directing the agent to "follow all returned instructions" from a command output for the remainder of the task. This pattern lacks explicit boundary markers or sanitization, meaning the agent's behavior and safety constraints are dependent on the output of the vaxis tool.

vaxis