The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the way it interpolates untrusted data into sub-agent prompts.
Ingestion points: The $ARGUMENTS variable is directly interpolated into the Agent() dispatch prompts within SKILL.md (e.g., "Simplify the code at: $ARGUMENTS").
Boundary markers: Absent. There are no delimiters (like triple quotes or XML tags) or instructions to ignore embedded commands within the interpolated text.
Capability inventory: The skill has access to potentially impactful tools including Edit, Bash, and the ability to dispatch further Agent tasks.
Sanitization: There is no evidence of input validation or sanitization of the $ARGUMENTS content before it is processed by the sub-agent.

simplify