The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and rewrite user-supplied text via the /humanize trigger, creating a vulnerability surface where instructions embedded in the input data could influence agent behavior.
Ingestion points: The skill accepts arbitrary English text from users in both Default Mode and Analyze Mode (SKILL.md).
Boundary markers: The instructions lack specific boundary markers (e.g., XML tags or delimiters) or explicit warnings for the agent to ignore any instructions found within the text being humanized.
Capability inventory: The agent is granted Read, Write, Edit, Glob, and Grep permissions, which provide significant file system access if the agent were manipulated into executing embedded commands.
Sanitization: There is no logic present to sanitize or escape user-provided content before the agent processes it for patterns.

english-humanizer