The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill functions by ingesting, analyzing, and rewriting text provided by users or read from files. This creates a vulnerability to indirect prompt injection if the processed content contains malicious instructions.
Ingestion points: Untrusted data enters the agent's context through user-pasted text via the /humanizer command and through files accessed via the Read, Grep, and Glob tools.
Boundary markers: The skill instructions lack explicit delimiters or "ignore instructions" directives to prevent the agent from accidentally executing commands embedded within the text it is tasked to humanize.
Capability inventory: The skill has access to Write and Edit tools, which could be misused to modify the local filesystem if the agent follows instructions hidden in the input text.
Sanitization: There are no instructions for sanitizing, escaping, or validating the external content before it is interpolated into the agent's rewriting process.

humanizer