The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill architecture is designed to ingest and process untrusted user input to generate prose, which introduces a surface for indirect prompt injection attacks.
Ingestion points: Data enters the context via the $ARGUMENTS variable and through several AskUserQuestion prompts (e.g., 'What's your take?', 'Personal experience?').
Boundary markers: The skill lacks explicit delimiters or specific instructions to the agent to treat the gathered 'substance' as data rather than instructions during the writing and review loops.
Capability inventory: The skill is capable of writing content to the local filesystem (draft-{topic-slug}.md) and invoking the writing-reviewer agent to analyze and modify these files.
Sanitization: There is no evidence of sanitization or escaping of the user-provided content before it is processed by the writing or reviewer sub-skills, although the filename itself is slugified.

write