The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text to remove AI writing patterns, which introduces a surface for indirect prompt injection.
Ingestion points: The skill ingests and processes text provided by the user for rewriting as described in SKILL.md.
Boundary markers: There are no explicit delimiters or "ignore embedded instructions" warnings defined to isolate the user's content from the agent's editing instructions.
Capability inventory: The skill uses Write and Edit tools to modify file content based on its analysis.
Sanitization: The instructions do not specify any sanitization or validation for the input text.
[EXTERNAL_DOWNLOADS]: The skill references external guidelines for identifying AI writing patterns.
Source: Refers to https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing in the references section of SKILL.md.
Note: This reference targets a well-known service for informational purposes.
[SAFE]: All identified behaviors, including the use of file manipulation tools and external references, are consistent with the skill's primary purpose as a prose editor and writing assistant.

humanize-writing