The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from several sources to build documentation artifacts and indices.\n- Ingestion points: The skill reads task files, PRD files, and codebase documentation in steps 2, 4, 7, and 9 of its behavior checklist.\n- Boundary markers: No specific delimiters or protective instructions are defined to distinguish ingested data from instructions.\n- Capability inventory: The skill utilizes file read and write operations, along with directory creation and deletion.\n- Sanitization: There is no evidence of content sanitization or validation before data from tasks or PRDs is interpolated into final documents.\n- [COMMAND_EXECUTION]: The skill instructs the agent to perform directory deletion as a final cleanup step after archiving. While intended for maintenance, this is a destructive operation that relies on variables like {timestamp} and {slug-work-item} derived from the local environment.

lcs-doc-finalizer