The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted data provided via arguments or external files and has the authority to modify the local file system using the Write and Edit tools.
Ingestion points: Data enters the skill's context through the $ARGUMENTS variable or via the Read tool when a file path is provided via the --file flag.
Capability inventory: The skill is configured with Write, Edit, Grep, Glob, and AskUserQuestion capabilities in its frontmatter, enabling it to persist changes based on the processed input.
Boundary markers: While the skill provides mechanisms to mask code blocks and blockquotes (--ignore-code, --ignore-quotes), it does not implement strict boundary markers or instructions to the agent to disregard commands embedded within the text being humanized.
Sanitization: The skill lacks explicit sanitization or validation procedures to prevent the agent from executing instructions that may be contained within the source material it is processing.

humanizer